Def COMMIT Demoboekje V2 LR

708723872936
80FUTURE6523
94OF7DATA243
18 495265
102957820146
84THE8BIG460THE BIG
FUTURE
OF DATA
4
Golden
demo’s of
COMMIT/

Colofon
The Board of COMMIT/
Arnold Smeulders
Peter Apers
Inald Lagendijk
Geleyn Meijer
Johan Vos
Produced by the office of
COMMIT/
Mieke van den Berg
Tjerk de Greef
Editor: Bennie Mols
Photos are provided by the
respective demo owners
Photo introduction:
Jan Willem Steenmeijer
Graphic design:
Studio Ron van Roon
Printer: Drukkerij Terts
© COMMIT/ October 2014
W
elcome to the world of Big Data. COMMIT/ is a public-private
research community for the advancement of Information
Technology in the Netherlands. The name COMMIT/ expresses
the dedication to communicate advanced IT-solutions to you. As
a public-private research community we are committed to pursue
the latest Information Technology and to present it to the Dutch society.
Running from 2010 – 2016, COMMIT/ is a precursor for the current
wave of Big Data, from advanced infrastructure, to extensive content
and web analysis, ad-hoc networks, data mining and to interaction.
We have collected fifty of our best results in demonstrators to make
Information Technology visible for you. Each demonstrator will illustrate
in what way it differs from the current state of the art.
Information Technology is the driver of economic innovation. Digital data
have entered everyone’s life, all professions, each business and all
relations. It is hard to imagine any such driver in the past (or it must be
electricity and the steam engine): the knowledge of Internet, the media
of the digital channels, the instant social media, the transformation of
the news, the organizational power of smart phones, the globalization
enabled by the network, the ease of interaction, the support of apps
which is starting to come and the statistics of professional live in Big
Data. Under whatever name, Information Technology has transformed
itself and the world around us overnight. Mobiles grew from zero to
general acceptance in ten years, not only here but all around the globe.
The same was true for smart phones, digital television and the Internet.
The impact of Big Data will follow with the impact on economics, use
and production.
The big, bright
future of data
3Introduction
Arnold Smeulders
Chair COMMIT/

The current fifty demo’s collected in this booklet are a direct result from
Dutch industry and have inspired IT-scientists of the Netherlands one-
to-one at the work table. As a result, the prototypes you are about to see
will be world-leading at least inside.
In science it is as important to be world-leading as it is in technology
business. In science there is an additional need to make well-motivated
and novel choices. In business there is an additional need to fit the
uniqueness into a robust business model. It is the strive for uniqueness
where science and technology business meet as the competitive edge
endures longer for unique solutions. Though not always easy, we have
put emphasis in our presentations to present the uniqueness of the
demo. You are invited to ask the question: Does Google already has
this?
Luckily, IT-science of the Netherlands is world-level, not just the network
but also the Big Data tools we are proud to present. Whether they are
local networks, information security, Twitter stream analysis, picture
analysis, interaction, sensor data processing, compute platforms or web
information management. You are invited to ask the question: Who else
does this world-wide?
Information Technology has this strange and almost unique impact
that it affects more than one, often completely different fields. For the
demonstrators, we have selected one application to make it real, but
almost always there is a second application hidden from a different field.
You are invited to ask the question: What else can you do with this?
Introduction 2 3Introduction
The fifty demo’s are sorted by type and by application. As a result,
the demo’s are indicative of what can be done with the results for the
purpose of inspiring you to other uses, maybe even your own use. We
would be more than thrilled if you are inspired to a different use.
The impact of Information Technology is phenomenal. Therefore, more
than any other discipline, all modern societies should have a high-level
of advanced Information Technology at hand. The impact is so big
because its reach is immediate, instantaneous, and far-reaching. The
impact is not limited to transporting bits from A to B, but expanding to
understanding the content of the information. Mastering the network and
the content leads to economic prosperity. IT is far too important to leave
the technology to foreign companies only as they reach our doors and
into our houses. Information Technology - and an active understanding
thereof - is a matter of national quintessence as is implemented in all
countries around us. A precursor to the topsectors, funding for COMMIT/
is for 45 percent from the government, 25 percent from universities and
academia, and 25 percent from industry and non-profit for a total of
110 M€ over six years. We are grateful for the vision which underlies
these choices.
On behalf of the COMMIT/ Board of Directors, you are invited to join in
the world of Information Techies!
Arnold Smeulders
Chair COMMIT/

Content 4 5Content
Legend
Volume & Velocity
Tracking the use of data all the way
To which music is the world listening?
Monitoring reputation on Twitter
Finding interesting information that you were not looking for
Geographically exploring Twitter hot-spots
Scavenger Hunt game helps in customer recommendations
Smart search using common sense
Heterogeneity & Quality
Finding new drugs by visualizing the effect of their ingredients
Rapidly finding variations between human genomes
Scan and print yourself in 3D
Exploration tool for investigative journalists
Automatic photo tag recommendation for Flickr
Finding the most interesting fragments in a tv-broadcast
MediaMill Semantic Video Search Engine
Visualization of topic evolution in news articles
Automatic risk assessment of vessels in maritime areas
Let an intelligent machine explain its decisions
Intelligent System Design
A facelift for old software
BTWorld: A Large-scale Experiment in Time-Based Analytics
Monitoring group emotions
Whiskers let your beloved ones feel your mood
Early validation of software designs
Emergency communication technology for crowd safety
Intelligent lampposts warn crowds in emergency situations
Scalability
Automatic testing improves software quality
Rapidly visualizing Wikipedia page views
Lost in your data? Let Blaeu give you a few tips
Predicting the earth’s climate with Graphics Processing Units
Building the Internet of the future
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
Demo 1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
Telehomecare that protects people’s privacy
Trusted healthcare services
Web-based tools for handling biomedical Big Data
How to get more out of food research
Wireless crowd monitoring in Arnhem
Using smartphones for emergency messaging
Measuring crowd densities for safety and efficiency
Physical User Interaction Styles
Tikkertje 2.0
Advising optimal step frequency for runners
Run, talk and don’t get injured
Intelligently grouping amateur runners
Learning with your body as interface
Playing with emotions in the Moodroom
Personalized interactive wall for elderly with dementia
Send your loved one a gentle caress via a tactile sleeve
An exercise game to reduce the risk of falling
A serious game for training social skills of police officers
Virtual learning helps people with language problems in
societal participation
Human Computation
Involving the end-user in the design process
Socially adaptive apps
Diabetes: Measuring Less to Feel More
Monitoring moods of workers reduces sick leaves
Coaching-app helps to find work-life balance
Quantifying your life for a better well-being
Molecular biobanks unravel the secrets of breast cancer
Enriching use and experience of radio content
Nichesourced annotation of cultural heritage collections
One-click semantic enrichment of scientific data
Evaluating the reliability of eHealth-data
Privacy transparency tools give more trust in eHealth
COMMIT/ projects and project leaders
66
68
70
72
74
76
78
80
82
84
86
88
90
92
94
96
98
100
102
104
106
108
110
112
114
116
118
120
122
124
126
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.

Each golden demo fits one of six ICT-
Science clusters, clearly respresented
by a color
Legend 6 7Legend
I
nteractive playgrounds are installations
that combine the benefits of traditional
playgrounds for children with advances
in technology. In our demo we show the
Interactive Tag Playground (ITP), a mod-
ern, interactive version of the century-old game
called ‘Tagging’ or ‘Tikkertje’ in Dutch.With the
Interactive Tag Playground we actively steer the
interaction between players. To this end, players
are being tracked using Kinects and additional
body-worn sensors. In its most basic version we
project differently colored circles around each
player to indicate the role of tagger or runner.
When these circles collide, a ‘tag’ is detected
and the roles of the players switch. We also add
novel interactive elements such as power-ups
and bonuses. Apart from entertainment, the Interactive Tag Playground is also a tool to study how
children interact with each other and with the environment. Our final aim is to automatically steer
the interactions in such a way that all players remain engaged and physically active.
ICT science question
How can we best track the players and understand their interactive behavior? What kind of interac-
tions are suitable at which moment during the play to maintain the engagement of players? How
can players be motivated to adopt certain play styles such as cooperation or competition?
Application
Our work can eventually be combined into playgrounds that allow for adapted play without any
restrictions such as the need to wear certain sensors. We see great potential for these playgrounds
in open spaces such as traditional playgrounds, shopping malls, sporting facilities and outdoor
squares.
37. Tikkertje 2.0
Robby van Delden
r.w.vandelden@utwente.nl
Alejandro Moreno
a.m.morenocelleri@utwente.nl
COMMIT/ project
IUALL Interaction for Universal Access
Physical User Interaction Styles 80 81
Alternative Application
Interactive play is everywhere. Not only children can benefit or enjoy interactive play. Previously, we
made an ambient interactive bar that was successful in entertaining adults. The techniques, both
for sensing and interaction, are largely the same. One can also think about offering play to people
who normally have limited play opportunities, such as heavily handicapped or demented people.
Interactive play might help in socializing, entertaining, triggering physical activity or even branding.
Nice to know
The behavior of adults and children while playing tag is largely the same. Except that children cheat
more.
In 2008, the UK Local Government Association promoted tag games stating that children are over-
protected (“wrapped in cotton wool”).[Wikipedia]
Variants of tag have fancy names such as “Cops and Robbers”, “Kiss Chase” and “Ringolevio”.
Quotes from participating students:
“This looks so cool!! I want to play!”
“I want to have this at home!”
“This should be a new sport, Olympic”
“I think this would fit very well in gyms or fairs”
An interactive whole-body game aimed at enhancing the player’s fun, physical and
social experience.
An interactive game environment that senses players’ behavior to facilitate a physi-
cally active, social and engaging experience with multiple other players without prior
calibration.
A novel game environment that mediates gameplay elements by sensing and tracking
players to promote targeted behavior or improving the game experience.
Tracking and analysis of human behavior to create engaging gameplay.
This research was supported by the Dutch national program COMMIT/
The icon represents a societal topic that
showcases the potential of each of the gold-
en demos. Four societal topics are chosen
in line with EU research programs:
Volume & Velocity addresses those
datasets that are huge but with the need
to be processed in real-time using smart
algorithms
Heterogeneity & Quality recognize
that raw data originates from a variety
of sources with different levels of quality
and focus on finding patterns that are
meaningful
Intelligent System Design centralizes
around the question how to create sys-
tems and standards that will continue to
work as the world evolves
Scalability closes the gap between
computational complexity and available
computing resources
explore natural interaction models to
interact with the physical world that is
engulfed with data
Human Computation allows intuitive
interaction with all sorts of data to bring
contextual knowledge and trust into the
process of understanding highly com-
plex data
Each golden demo comes with a short
statement explaining the golden demo
to one of the following personas:
General Public: explains what
is in it for companies or society
at large
Product Developer: describes
to an entrepreneurial youngster
the innovation potential and
societal impact
Innovation Manager: provides
arguments at the boardroom
level for a trial project
Scientist: reveals to academics
the ICT-science challenge
underlying the golden demo
These researchers are responsible for the
golden demo
Partners that are involved with the golden
demo

D
ata analysis and transformation are in-
creasingly important activities in both
scientific research (e.g. climatology) and
other fields (e.g. open government data).
Unfortunately it is hard to assess the
trustworthiness and quality of the results with-
out knowledge of what data the outcome was
based on, and through what procedure the out-
come was reached. This information about enti-
ties, activities and people involved in using data
is called data provenance.
Our demo shows the integration of data prov-
enance tracking and visualization in an existing,
popular data science environment. The demo is an application of our work based on the PROV
W3C standard, provenance visualization and tracking.
Our work allows for fine-grained tracing of conclusions in scientific papers to intermediate results,
other publications, across applications and source data.
Data are manipulated in a wide variety of tools. It is a grand scientific challenge to construct, recon-
struct, communicate and connect data provenance traces. In solving this challenge we have to deal
with a lack of standards and integration in tools. Another challenge is to integrate data provenance
in environments that scientists already use, without forcing them to learn a new tool or adjust their
way of working.
Application
We apply our technology to support data scientists in creating a better argumentation for their
research outcomes. We integrate the PROV W3C standard as part of an existing, widely used
1. Tracking the use of data
all the way
Rinke Hoekstra
rinke.hoekstra@vu.nl,
www.data2semantics.org
COMMIT/ project
Data2Semantics From Data to Semantics for Scientific Data Publishers
Volume & Velocity 8 9
open source data science environment (IPython Notebook) as well as version control systems and
personal file storage solutions.
Currently provenance tracking has only been implemented in highly controlled, closed environ-
ments such as scientific workflow systems (WINGS, Taverna). Our innovative visualization tool
PROV-O-Viz visualizes the flow of information through the provenance graph, giving users better
insight in the important aspects of their workflow.
Provenance tracking has wide application areas outside scientific research. In the context of Big
Data in industry and government, it becomes increasingly important to know the origin of individual
datasets. This is not only because of reliability and trust issues, but also because of legal reasons
such as license compatibility, copyright, intellectual property right and privacy.
Nice to know
Without sufficient provenance information, scientific research cannot be reliably reproduced. Phar-
maceutical company Bayer halts about two-thirds of drug target-validation projects, because ex-
perimental findings reported in literature cannot be reproduced.
Money well spent? Provenance is key in improving the efficiency, reproducibility, integ-
rity and trustworthiness of research.
Anyone can publish dead data. But can you publish it in a way that others can find it,
combine and reuse it?
Innovations should be seamlessly integrated in everyday practice, with a maximum ef-
fect on the quality and traceability of information exchange.
Suppose a reviewer walks up to you and says: “That number in Table 1, where does
that come from?” Well, do you have an answer?
Volume & Velocity

W
e have developed the web application
Streamwatchr, that monitors Twitter to
find out to which music people are lis-
tening. Streamwatchr offers real-time
insights into music listening behaviour
around the world.
Using the button ‘Now’, we show the stream of
plays coming in. It can be great to discover and
listen to some unknown bands that others are
listening to. Using the button ‘Hot’, we show a
real-time chart, based on the current popular-
ity of songs and artists. Finally, using the button ‘Unexpected’, we try to find songs and artists
that could be booming soon. Streamwatchr’s engine for interpreting music listening behaviour on
Twitter uses YouTube, Musicbrainz, last.fm, lyricsNmusic for mapping and analyzing an incoming
stream of tweets: six tweets per second, five hundred thousand tweets per day.
The core scientific challenge that Streamwatchr addresses is how to interpret highly dynamic user-
generated texts. Streamwatchr maps the content of text messages to a knowledge base in real-time.
Application
Streamwatchr is a consumer-oriented web application based on tweets about music. The interface
shows the music that is being played and listened to around the world in real-time via flipping tiles.
By clicking a tile, users can play songs via Youtube and find related songs via Streamwatchr’s
recommender system. Users can also see what people are singing along to now, or which parts of
the lyrics are most popular for a song. Streamwatchr’s private partners are 904Labs and Eagerly
Internet. Due to the highly competitive arena in consumer-oriented music discovery (e.g., iTunes,
Spotify, Milk), Streamwatchr will use its backbone technology for developing business-to-busi-
2. To which music is the world
listening?
Maarten de Rijke
derijke@uva.nl
streamwatchr.com
COMMIT/ project
INFINITI Information retrieval for Information services
ness products different from the consumer-oriented web application. A non-profit partner can be
Musicbrainz for using Streamwatchr’s song and artist popularity signals. A commercial partner can
be Spotify, which can extend their signals with those from Twitter using Streamwatchr’s technology.
For these applications, the core technology is in place but APIs for broadcasting the data are still
to be developed.
Unstructured text analysis is broadly applicable. The technology behind Streamwatchr can be ap-
plied to analyze behavioural patterns around other consumer-oriented products, like movies, tv
shows, food, or wish lists to name a few. Our technology allows for generating real-time popularity
charts, detecting anomalies as early signals of the next music hit, and radio functionality via a
recommender system that is updated in real-time with every single tweet.
Nice to know
People around the globe report about their music listening behaviour, in half a million tweets. Every
day. Streamwatchr encounters between two hundred and three hundred new bands per day. Every
day. For its semantic interpretation engine, Streamwatchr fires two hundred million queries per
month. Every month.
Streamwatchr’s technology transforms the content from social media into meaningful
signals. Streamwatchr has been selected as one of 12 projects that got funding for
SXSW, the major event for music industry, and attracted very positive feedback.
The technology behind Streamwatchr adds a layer of intelligence in products that work
on behavioral signals. Potential appications include online reputation management,
discovery and predictive applications, and recommender systems.
Activities shared online is a powerful but hard-to-capture signal due to their high vol-
ume and their unstructured nature. Streamwatchr marks the first step beyond tedious
manual annotation of these data by capturing their essence automatically.
To understand the world as it happens, our technology the mapping of unedited text
to knowledge bases to see patterns in human behavior. Entity recognition and disam-
biguation is particularly challenging for songs and artists in the long-tail.
Volume & Velocity

I
n an online society reputation manage-
ment is essential. Tweets can quickly
make or break the reputation of brands,
companies and organizations. We pre-
sent a semi-automatic tool designed to
monitor the reputation of companies, organiza-
tions and people on Twitter. The tool is called
ORMA: Online Reputation Monitoring Assistant.
ORMA automatically annotates tweets for
relevance and reputation: Is the tweet about the entity? Has the tweet positive or nega-
tive implications for the reputation of the entity? The tool also groups tweets by top-
ics and displays topics in decreasing order of relevance from a reputational perspective.
ORMA helps the user to understand the contents being analyzed and to produce a manually an-
notated version of the data, using the output of the automatic annotation processes.
How can the annotation process of unstructured texts be made more efficient? To solve this scien-
tific challenge, we develop new techniques based on advanced semantic information retrieval and
natural language processing.
Application
Our reputation monitoring tool assists experts by suggesting automatic annotations that can be
manually changed. Preliminary results suggest significant time saving compared with not using au-
tomatic annotations. We have developed the tool in collaboration with the leading Public Relations
consultancy firm Llorente & Cuenca.
There are already reputation monitoring tools available on the market, for ex-ample Brandwatch,
MYSPYSPOT, Talkwalker and MetricPoint. However, ac-cording to a recent survey by Llorente &
3. Monitoring reputation on
Twitter
Cuenca, they do not satisfy the necessities of reputation experts due to a lack of relevant function-
alities and poor performance.
Our research can be applied in other contexts involving consumer portals, opinions portals, news
aggregators, marketing strategies, etc. New applications possibly require different visualization
solutions, but the techniques we use for natural language processing and machine learning will be
generally applicable.
Nice to know
ORMA has already been used in a real annotation process in order to build a corpus of tweets for
developing and testing Online Reputation Management algorithms at RepLab 2013. This is a com-
petitive evaluation exercise for Online Reputation Management systems.
Quote
“Using this tool the reputational experts will improve their efficiency aiming to analyze more data
with the same time and improving the quality of their re-ports.” − Analyst of consultancy firm
This tool can be effectively used by politicians to get powerful and comprehensive
insights on their online reputation as distilled from relevant Twitter posts.
ORMA integrates a number of algorithms into a reputation monitoring solution that
lies between full automation and manual analysis, combining automatically generated
results and experts’ feedback.
We propose a unique reputation monitoring tool able to learn from manual annotations
of tweets to provide reputation insights in a more efficient and accurate way.
ORMA features advanced techniques for filtering, polarity classification of tweets, topic
detection and their ranking by priority from a reputation management perspective.
This demo is part of the LiMoSINe project in close cooperation with the COMMIT/ demo to which music is the world listening?
(Nr. 2, also known as the StreamWatchr demo). The LiMoSINe project runs within the European Community’s Seventh Framework
Programme (FP7/2007-2013) under grant agreement nr 288024.
Volume & Velocity
Julio Gonzalo
julio@lsi.uned.es
Jorge Carrillo de Albornoz
nlp.uned.es/orma/
www.youtube.com/watch?v=ZSYVceckrrc
LiMoSINe Project
limosine-project.eu/

W
hen you use a search engine like Google
or Yahoo, you are basically looking for an
answer to a specific question. However, in
everyday life you sometimes stumble upon
interesting information that you were not
looking for, for example when reading a news-
paper or jumping from hyperlink to hyperlink on
the web.This surprise effect is called serendipity.
We are building a new type of search and ex-
ploration engine that promotes serendipity. Our
serendipitous search engine, called DEESSE,
provides results that are both relevant to
the user’s current interest, and interesting, to encourage the user to continue the exploration.
DEESSE takes its data from both two major sources of user-generated content: Wikipedia, a col-
laboratively edited encyclopaedia, and Yahoo Answers, one of the most popular question-answer-
ing platforms where users freely exchange opinion and advice
Serendipity and interestingness are subjective notions. Can we, despite this subjectivity, find a sci-
entific approach to define, operationalize and evaluate what makes a search result serendipitous?
No previous research or application has fully succeeded in this yet, although some attempts have
been made to introduce serendipity into browsing social networks and portals.
Additional challenges: How do we analyze data collections at a very large scale? How do we ex-
tract interesting and meaningful information from unstructured text?
Application
We have built a fully functional prototype of a serendipitous search and exploration engine that
uses multiple languages. Using Wikipedia and Yahoo Answers the engine extracts a network of
related entities. Furthermore, it uses metadata about writing quality, topical category and sentiment
(emotions evoked by the text: from very negative to very positive).
Although recent research has focused on serendipitous search, few products have been developed
that enable users to explore data using this paradigm. A tool like DEESSE is a direct competitor of
existing search engines (Google, Bing, Yahoo) and other exploratory tools.
Search engines are a key technology. To make them function better has a wide impact.
Nice to know
Do you know that the Yahoo Answers entity network is extracted from 67 million questions and 261
million answers?
Assisting users in unexpected but pleasant and meaningful discoveries when search-
ing.
As user engagement is crucial for a successful product, we devise new ways of catch-
ing the curiosity of users by immersing them in a serendipitous search experience.
Using serendipitous search, we provide interesting and relevant facts that answer or
complement the user’s information need, that are otherwise buried in the information
overload on social media nowadays.
Driving serendipitous discoveries in search by exploiting large scale entity networks
from social media.
Volume & Velocity
4. Finding interesting information
that you were not looking for
Ilaria Bordino
bordino@yahoo-inc.com
Olivier Van Laere
vanlaere@yahoo-inc.com
deesse.limosine-project.eu/
LiMoSINe Project

G
eographical data are typically visualized
using various information layers that are
displayed over a map. Interactive explo-
ration by zooming and panning actions
needs real-time re-calculation. For lay-
ers containing aggregated information (such
as counts and sums) derived from voluminous
data sets, such real-time exploration is impos-
sible using standard database technology. Cal-
culations require too much time.
We have developed database technology that
accurately aggregates data so that they can geo-
graphically be explored in real time. The technology is a plug-in to common open source technology.
We demonstrate our technology by the exploration of tweeting hot-spots based on twenty to thirty
million geo-tagged tweets from The Netherlands and the UK.
A common operation in calculating with multidimensional data is the computation of aggregates.
In order to obtain exact results with high performance from high data volumes, we face the chal-
lenge of finding clever ways of pre-calculating data as much as possible. An additional technical
challenge is to develop technology that fits into standard open source database and GIS software.
Application
In geo data visualization, the ability to quickly develop new information layers is important. Although
many solutions exist, there is a niche: the combination of visualizing aggregation information, inter-
active data exploration in real-time, Big Data, calculating exact numbers instead of approximations,
and doing so with common open source technology. Our technology for the first time integrates all
these features.
5. Geographically exploring
Twitter hot-spots
Andreas Wombacher
andreas.wombacher@nspyre.nl
Jan Flokstra
Jan.Flokstra@utwente.nl
www.utwente.nl/ewi/db/research/currentprojects/Timetrails/
COMMIT/ project
TimeTrails Spatiotemporal Data Warehouses for Trajectory Exploitation
Our research partners are the companies Arcadis and Nspyre. They both have struggled with this
combination of requirements in many of their projects.
Our database index technology is not specific to geographical data. It can be used with all types of
multidimensional data. Visualization in business intelligence or eScience can also benefit from it.
Nice to know
The DCMR Milieudienst Rijnmond has used our technology to investigate whether people send
tweets about unpleasant odours as a possible signal of danger. This turns out not to be the case,
probably because people think that nobody reads the tweets anyway. But if people have the idea
that their complaining tweets are read, then tweets might be much more convenient than the report-
ing of unpleasant odours by telephone.
Interactively explore data on a map: showcase tweeting hot-spots.
Interactively explore data on a map: showcase tweeting hot-spots and the dialogue
with the public through social media.
Interactively explore data on a map: supporting the dialogue with the public through
social media using common open source technology only.
Database indexing of precise aggregations on all offsets and zoom levels necessary for
interactive exploration of spatio-temporal and other multidimensional data.
Volume & Velocity

T
here are many web portals aiming to rec-
ommend businesses like shops, restau-
rants and cafes. They are interested to
know how many people visit them. We
help them by developing algorithms for
determining actual visits based on GPS-traces
from mobile phones and public data from the
web only.
For proper scientific validation of the developed
techniques, a large data set with GPS-traces is
needed that is truthfully labelled with actual vis-
its to points-of-interest. In our demo we show a
means to obtain such data, namely a mobile Scavenger Hunt game (‘Speurtocht’). The game has
been used for the Kick-In events of the University of Twente in August 2013 and 2014.
What is the best way to compute a visit to a point-of-interest based on GPS-traces from mobile
phones?
Such actual visits can be computed from geographic data as an intersection of the GPS-trajectory
with a polygon describing the circumference of the point-of-interest. Polygon-data are, however,
not available. We have developed algorithms for estimating circumference polygons of point-of-
interest-objects by analyzing more coarse-grained map data and data on other objects.
Our algorithms produce quite accurate results even when data of substandard quality is used. The
latter is important, because it allows the application to use only publicly available data.
Application
For the commercial recommendation of items, one needs indicators of suitability per target group
6. Scavenger Hunt game helps in
customer recommendations
Victor de Graaff
v.degraaff@utwente.nl
Maurice van Keulen
keulen@cs.utwente.nl
https://github.com/utwente-db
COMMIT/ project
and other categorizations as expressed by points-of-interest (a shop, a restaurant, or a cafe). Our
partner EuroCottage would like to recommend holiday homes taking into account neighbouring
points-of-interest and indicators such as popularity. We automatically construct holiday home pro-
files with data on neighbouring points-of-interest. Factual data are harvested from the web. Popu-
larity is derived from analyzing GPS-traces from mobile phones of previous vacationers.
Determining actual visits to places can provide important information in many different domains.
Cars, planes and animals all produce trajectories. For example, automatically determining the feed-
ing and nesting places of animals from sightings or GPS-tags helps to understand animal behav-
iour.
Nice to know
Two applications have been created on this platform already: Kick-In Quest (2013) and Kick-In
Scavenger Hunt (2014). They are available for download in the Play Market and the App Store.
Going beyond social media ‘likes’ while your phone keeps track which shops, restau-
rants and cafes you visit for the purpose of recommending holiday destinations.
In geographic information systems analysis, from your phone data it is determined
which shops, restaurants and cafes you visit.
The system demonstrates the possibilities of intelligent social media supported maps.
Given trajectories and estimated polygons for points of interest, accurately determine
true visits; and how to obtain truthfully labelled GPS-data for validation using a mobile
game.
Volume & Velocity

S
earching information on a specific topic
can be time consuming. Instead of using
brute force searching that does not use
any knowledge about the world, we de-
velop smart methods that use background
knowledge on specific domains. This back-
ground knowledge is formalized in an ontology:
a network of concepts within a domain. An on-
tology in the food-domain might for example for-
malize that ‘Jonagold’ is a type of ‘apple’, which
is related to ‘orchard’.
We have developed two tools that enable smart search: the ROC+ tool and the SIEVE tool. ROC+
helps a domain expert to create his own ontology by making associations, using publicly available
vocabularies. The SIEVE tool in turn uses this ontology to refine a collection of documents for ex-
ample in a library, a company-specific corpus or the internet.
The ontology and the refined document set enable end users to efficiently find or organize specific
and high quality information.
Although domain-specific ontologies enable smart search, their construction is hard and time-con-
suming. Domain experts are generally required to distinguish relevant from irrelevant concepts.
The scientific challenge is to minimize the effort by experts while maximising its usefulness.
Application
The combination of our two tools minimize the effort needed from experts to build a specific docu-
ment set and a dedicated ontology: the time needed can be reduced from weeks to hours.
On our FoodVoc-website we publish ontologies related to the food domain that can be used by
these tools. One application is the Valerie project, which generates innovation advice for European
7. Smart search using
common sense
farmers based on results from EU-projects. The method has also been applied to assist in identify-
ing innovation opportunities for small and medium food enterprises.
Our two tools can be used in any other domain than food, for example pharmaceutics, chemistry,
finance or automotive. With our experience in the food domain we can organize ontology creation
sessions and assist in handling document repositories.
Nice to know
Users have experienced our ROC+ tool also as a team building tool.
Quote
“Your tools help us to retrieve relevant information tailored to our customers’ needs” − Ivo van Ham,
innovation manager of Syntens about sCore, a project that has used ROC+
Smart search tools to improve the innovation process.
Fun tools for smart search kick start our innovation process.
Smart search puts information at our employees’ fingertips, so that assets previously
wasted on ineffective search are put to good use elsewhere.
With smart search I find high-quality information for my research.
Volume & Velocity
Nicole Koenderink
Nicole.Koenderink@wur.nl
COMMIT/ project
e-FOOD e-Foodlab

F
inding new drugs to cure diseases is a
hard task. This is because the chemicals
in the drug interact in a very complex way
with the cells and proteins in the human
body. Visualizing this complex network of
interactions is important to improve the devel-
opment of new drugs.
Our demo shows how the interaction between
the chemicals in a drug and the proteins in the
body can be interactively explored in a rapid way. This rapid interaction makes it possible to get
answers while you think, as opposed to waiting for answers, which breaks the train of thought.
How to effectively visualize large graphs? This is a hard problem. Graphs that contain more than a
thousand nodes tend to become cluttered using most visualization algorithms.
The implementation we show in this demo accelerates the visualization by using a graphical pro-
cessing unit (GPU). This makes it possible to interact with large graphs (in the order of magnitude
of one million nodes).
Application
Our demo is an application of a well-known visual analytics tool (SynerScope) to the pharmaceuti-
cal data from the international Open PHACTS-project. Our visual tool combines a number of differ-
ent visualization techniques to achieve highly scalable graph visualization while taking into account
the interactions that help the user.
Natural competitors are visual analytics tool suites, like QlikView, Spotfire and Tableau, and open
source tools like Gephi. However, they all miss scalable network visualization and rapid interaction
response time.
8. Finding new drugs by visualizing
the effect of their ingredients
Heterogeneity & Quality 22 23
The SynerScope visual analytics tools have been used in forensics, fraud detection, anti money
laundering, risk-based pricing of insurance products, insurance claims process optimization, and
e-mail communication analysis.
Future high potential markets are: smart grid monitoring, logistics optimization, cybersecurity, tele-
communication network monitoring, high-performance cluster maintenance, auditing of supply
chains.
Nice to know
Usability studies have shown that SynerScope can be used effectively by high school students after
a tutorial of only thirty minutes.
Quote
“SynerScope, this is an amazing story, they are solving very complex issues by looking at billions
of transactions, […] They are saving the world of their share of the two hundred billion dollars that’s
lost to fraud annually.” — Bill McDermott (CEO SAP)
Big data does not change the world, insight in big data changes the world. Large-scale
visualization provides this insight.
Interactive exploration of network data reveals the hidden drivers behind trends. Know-
ing these drivers opens up new business opportunities.
Interactive visual analytics turns data analysis upside-down. We don’t just answer
questions, we also question answers.
Explore your research data in completely new ways and discover patterns you didn’t
know existed.
Willem Robert van Hage
willem.van.hage@synerscope.com
www.synerscope.com
COMMIT/ project
Open PHACTS consortium

W
e have developed a software tool, called
the Variant Description Extractor, that rap-
idly compares one human genome with
another in order to find small but crucial
genetic differences. Our tool generates a
complete description for the human genome in
about four hours.
The human genome contains twenty to twen-
ty-five thousand genes distributed over a long
molecule, called DNA. Genes can be described
by long strings of the four letters A, C, G and T. Each of them stands for a simpler molecule in the
DNA. On average, humans only differ 0,1% genetically from each other. However, especially for
finding causes and solutions to diseases it is crucial to find and understand these small differences.
The main scientific challenge is twofold. First, how to calculate short and unique descriptions from
long strings of the letters that compose the genes? The genes can hold thousand to many millions
of these four letters.
Second, how can this calculation be done within an acceptable and minimal amount of computa-
tional time?
Application
Our Variation Description Extractor is integrated in the Mutalyzer suite. Mutalyzer is a very popular web-
based software tool primarily designed to check descriptions of sequence variants according to the
standard human sequence variant nomenclature of the Human Genome Sequence Variation Society.
Mutalyzer aims to encourage the proper use of nomenclature in publications and reduce redun-
dancy in gene variant databases. This greatly improves the findability of variants. Ultimately, these
genetic variant descriptions are used in the diagnostics of hereditary diseases.
9. Rapidly finding variations
between human genomes
Jonathan K. Vis
jvis@liacs.nl
www.mutalyzer.nl
COMMIT/ project
Although our software tool is now focused on comparing DNA-strings composed of four letters, it
can easily be used on strings composed of other signs. The tool might for example be applied to
describe the differences between two natural language texts. The fundaments of the tool and the
algorithm do not have to be changed. Certain operators that are now especially designed for DNA-
comparison can easily be disabled.
Nice to know
The Variant Description Extractor has already been downloaded 66 million times.
The Mutalyzer suite enables medical researchers to find all variants and combina-
tions of variants in DNA sequences quickly. These variants can be used to effectively
research complex diseases.
We use a state-of-the-art algorithm to quickly find variants in DNA sequences. The al-
gorithm is supported by a Python suite of compatible tools adhering to the well-known
HGVS standard.
The efficient extraction of variants on a individual level enables the targeting of dis-
eases on a person-to-person basis. Making it a prerequisite for and a first step towards
personalized medicine.
We propose an efficient algorithm for the extraction of biological meaningful descrip-
tions of the variance between complete genome sequences with minimal length and
computation time.

T
he last few years have seen an enormous
development in 3D-printing. Consumers
can now buy their own printers for use
at home. However, if you also want to do
your own 3D-scanning of the object you
want to print, present laser scanners are far too
expensive. We have solved this problem by us-
ing a low resolution, low cost depth camera.
In our demo we show the 3D-scanning of ob-
jects and scenes, such as faces, bodies, furni-
ture and rooms. To obtain a full 3D body scan,
a person stands still on a platform. In less than
half minute we perform a 360-degrees full body
scan. As a take home gift, you will receive your
own 3D scanned or printed body model.
The main challenge is how to use a low resolution, low cost depth camera to perform real-time 3D-
scanning in order to reliably print a 3D-model. The present high resolution 3D-laser scanners cost
about twenty thousand euro. These scanners are not affordable by consumers and they are not user
friendly. Moreover, the necessity of post-processing makes them less applicable for daily usage.
An important sub-challenge is to develop new 3D-modeling solutions that can be used on the con-
sumer market. Like with the scanners, the existing modeling solutions are often too expensive and
too complex for consumers.
Application
We provide a user-friendly and real-time 3D-object scanning solution at a very low cost: about
ninety euro. As a consumer you only need our software, a standard computer and a depth sensor
in order to scan real world scenes such as faces, bodies, furniture, rooms, cars, art, etcetera. Our
solution will help to reduce the gap between 3D-modeling and 3D-printing.
10. Scan and print yourself in 3D
Theo Gevers
Th.Gevers@uva.nl
3duniversum.com
COMMIT/ project
3D-printing is broadly applicable and 3D-scanning is an important technology to that purpose.
3D-scanning also offers new business opportunities for real estate agencies, for custom made
clothing (eye wear, dresses…) and for the movie industry. 3D-Recording also allows applications
in health monitoring and sports. Moreover, a large collection of 3D-scanned real world objects will
help the field of artificial intelligence, in particular automatic image recognition.
Nice to know
With our software you can create a 3D-selfie in less than a minute.
You do not have to invest thousands of euros on 3D scanners or you do not have to
spend time to learn complicated 3D CAD systems. This demo provides you a low-cost
and real-time 3D scanning solution.
Revolutionize the way you interact with objects such as bodies, rooms, and furniture.
They are all in 3D now.
You can scan, visualize, interact and share real-world objects in 3D with a low-cost
camera.
Real-time 3D reconstruction and recognition.

G
iven a pile of documents, a journalist faces
the task to dig deeply into the various lay-
ers that could yield a newsworthy story on
political and other powers. Especially for
investigative journalists we have devel-
oped the exploration tool FIDO (Freedom of In-
formation Document Overview). FIDO supports
the exploration, search and analysis of large
collections of texts, such as the ones that jour-
nalists can get through Freedom of Information
Act-requests (WOB in Dutch).
The journalists’ starting point is the Documentcloud.org framework, where WOB-documents are
stored, processed and annotated. FIDO extends this tool suite in order to support search for enti-
ties and context more intelligently. It also produces powerful visualizations of social networks, word
clouds, maps and timelines.
How can we discover in a large set of documents the key entities? How can we find the evidence
central to these entities and the contextualization of mentions of the same entities in other sources?
Specific technical challenges are the heterogeneity of resources to be integrated, the multi-lingual
character inherent to the real-world setting and the noise resulting from optical character recognition
(OCR - used to automatically generate a digital text by scanning an analog, printed text). Unique
about FIDO is that it tackles these issues ‘on-the-fly’, instead of via a priori analysis.
Application
Our demo shows how FIDO supports example scenarios handed in by journalists. One is inspired
by an investigation of the Dutch-Swiss company Vitol, the largest independent energy trader in the
world. Largely unknown to the public, Vitol emerges as a spider in the political-economic web in
11. Exploration tool for
investigative journalists
Matje van de Camp
infiniti.taalmonsters.nl/
COMMIT/ project
FIDO. Compared to how the initial article on Vitol was written, FIDO offers a much more effective
workflow.
The analysis of piles of documents is useful in many places.
The type of challenge faced by investigative journalists can be likened to other intelligence activities
in which the starting condition is ‘a pile of documents’. Therefore, alternative application areas are
business intelligence, police investigation and academic research. We have plans to do usability
studies with financial analysts responsible for reputation and trend analyses at a big investment
management company. FIDO could increase the effectiveness of text analytics and show its rel-
evance for generating financial forecasts.
Nice to know
The development of the demo followed an interactive user-centric design process. Investigative
and data journalists were and are being interviewed on their workflow and how FIDO should be
shaped to fit this workflow.
Investigative journalists dig through more and more data to give the public what it
wants: freedom of information.
Semantic search: integrating the most robust natural language and search technolo-
gies for unlocking information from text at superhuman scale and speed.
The only way to keep up with companies and governments producing increasing
amounts of textual data, is to move to next-generation semantic search technologies.
Semanticizing and linking heterogeneous, multilingual sources with state-of-the-art
natural language processing and information retrieval technologies.

P
eople from all over the world increasingly
share their knowledge, experiences, opin-
ions, photos and videos online. New kinds
of language-based search technologies
are needed to make this multi-media con-
tent effectively searchable.
In our demonstration, we offer a web interface
which allows the user to annotate images taken
at large scale social events: i.e. to describe in
words some extra information about the image.
Annotation is done in order to make huge collections of photos better searchable by keywords.
We use social media data, as well as techniques from information retrieval and image processing,
in order to automatically recommend photo tags. This makes the annotation process much easier.
A photo sharing web service such as Flickr can benefit from our automatic photo tag recommenda-
tion.
Automatically detecting objects and scenes from images is one of the biggest challenges in arti-
ficial intelligence. This is due to the fact that computers have no understanding of the world (the
semantic gap).
Previous methods have used either the information of the pixels of the photos, or the tags added by
the user in order to automatically annotate images. We offer a new solution which combines visual
appearance, context and social media data in order to offer tag suggestions for photos.
Application
Our demonstration applies to the process of annotating images uploaded online which are often
annotated with less than four tags. Such a small amount of tags makes photo retrieval and recom-
Philip McParlane
p.mcparlane.1@research.gla.ac.uk
tttagged.com/demo/
LiMoSIne Project
mendation difficult. Our work could for example be applied in an image sharing website domain
such as Flickr.
Annotation is broadly applied to build knowledge bases.
Our application is best suited for image annotation but could alternatively be applied to the annota-
tion of any web object, such as videos, news articles, web documents etc.
Nice to know
With the high velocity of social media, we are able to mine the ‘wisdom of the crowd’ to collect rel-
evant tags describing images taken at large scale events as they happen.
Quote
“Using the keyboard to tag images is slow and boring. This demo instead makes the whole process
possible using only the mouse.”
Make image tagging easy by using context and content recognition.
Exploit historical big data in order to compute on-the-fly tagging recommendations in a
image sharing website scenario.
Offer an easy image tagging scenario to the user by exploiting big data.
Combine user tags, textual and visual features in order to suggest the best tags for a
photo.
12. Automatic photo tag
recommendation for Flickr

W
e have developed SocialZap, a multimedia
search engine that finds the most interest-
ing fragments (‘zap points’) in a television
broadcast, based on microblog posts like
tweets and socially tagged photos.
The main novelty of SocialZap is the fully-au-
tomatic transfer of the learned viewer’s interest
from textual posts to the visual channel. There
is no need for any manual effort in the process.
Once SocialZap finds the zap points, users can easily browse through a television broadcast and
directly watch the interesting fragments. Thus, SocialZap adds social experience to watching televi-
sion.
What happens where in digital video? The fundamental problem in video retrieval is that computers
can − at present − only extract low-level features from a video signal, whereas humans interpret
the data in a high-level conceptual way. It’s a grand scientific challenge to bridge this so called
‘semantic gap’. In particular we consider here the synchronization of audio and text signals when
they refer to one and the same event.
Application
The questions we pose here are highly relevant in a world in which visual communication is ever
more important.
Existing web services like UitzendingGemist allow viewers to watch missed television broadcast
on the web. However, a system that directly suggests the most interesting fragments to watch,
based on social media, is non-existing. The particular challenge we face is the temporal mismatch
between the moment that the user tweets about a concept and the moment at which it appears in
the television broadcast. The tweet-time can radically differ from the appearance-time as viewers
Cees Snoek
cgmsnoek@uva.nl
sealincmedia.wordpress.com
COMMIT/ project
SEALINCMedia Socially-enriched access to linked cultural media
either anticipate appearances or continue to tweet about topics that have previously appeared.
In our SocialZap demo, we analyze data from social media to suggest interesting concepts (zap
points) in television broadcast. SocialZap uses textual information from Twitter posts related to a
television broadcast of interest, which provide a rich source of information of what viewers find
interesting.
Video is everywhere and increasing, too much to watch all.
Our technology might be used in the digital humanities or by intelligence and security services.
Nice to know
SocialZap provides the only service in the world that automatically links social multimedia
noteworthiness to television content.
SocialZap has won the US-run NIST TRECVID video search engine benchmark in 2013 and
2014.
The things people say about TV are interesting and valuable: SocialZap allows that
value to be exploited by the viewers of catch up TV, by linking micro-blog posts directly
to interesting moments.
Both social media and TV are important sources of entertainment and information, but
they live in separate worlds. SocialZap brings them together to enhance the appeal and
usefulness of catch-up TV.
With catch-up TV we never miss anything on TV. SocialZap is the first to take it one
step further, by offering users particularly noteworthy time points, reducing the time
needed to catch up.
Computer vision has matured to the degree that it is useful in consumer applications. In
SocialZap, new algorithms provide jump-in points to viewers by exploiting information
from social media.
13. Finding the most interesting
fragments in a tv-broadcast
Auxilium

T
he demo shows the MediaMill semantic
video search engine, a system facilitating
access to the content of large collections
of video streams. The system is based on
a large lexicon of visual concept detectors,
complex event detectors and an interactive vid-
eo browser learned from example video mate-
rial. MediaMill provides access to the content of
videos without the need to label each and every
video. The old way was to let someone tell you
what is in the video, like is the common prac-
tice in YouTube. In order to make progress the
power of the human eye to tell in words what
is seen needs to be reproduced. MediaMill has found a way to do that by machine learned image
recognition technology translating the pixels of an unknown video into a descriptive text.
The world is full of digital videos and images. In this deluge of visual information, the grand chal-
lenge for computer vision is to unlock its content directly by understanding what is in the image.
Application
Digital images containing ‘busses’, ‘beaches’ or ‘babies’, or any other concept for which a visual
representation can be made like ‘kitchen, even as abstract as ‘wedding’ are being identified. The
latest addition to recognize video fragments on the basis of events like ‘making a purchase’, ‘doing
a magic trick’, or ‘rock climbing’. None of the recognition is perfect, and this will not be for some time
to come, but the techniques are helpful in many interactive applications.
Cees Snoek
cgmsnoek@uva.nl
www.mediamill.nl
COMMIT/ project
As digital images are everywhere, access to the content of images is broadly needed.
The MediaMill video search engine can be applied for automated moderation of social media,
interactive search in forensics, home video handling, image sorting for real-estate brokers and
other professionals, industry product maintenenance, media publishers, and more where there are
lots of digital images or videos.
Nice to know
The software has been transferred to Euvision Technologies, a University of Amsterdam spin-off.
Quote
The University of Amsterdam has landed in the top-three of the open-innovation TRECvid inter-
national competition run by the US-bureau for standards NIST on image and video search for 10
years in a row, amidst fierce competition from international universities and companies.
In addition, Cees Snoek receives Netherlands Prize for ICT Research 2012
NWO, 24 juli 2012 and Video search engine matures, NWO, September 2013.
The world is nowadays full of digital images - industry, video, web, professional, busi-
ness or monitor. The computer will sort what is going on.
In the world-wide open-innovation competition of 50 image sort engines - with IBM,
Oxford and CMU - this product has been in the top 3 for the last 10 years.
A camera plus automatic categorization learned from example images is efficient in
many applications: social media, trend, monitoring, mobile, photo-memory.
The dominant human sensor, the eye, was weakly developed for computers. Naming
things, actions, and states in the picture is a huge cognitive step.
14. MediaMill Semantic Video
Search Engine

M
ore and more texts are available in digital
form. Especially when these texts go back
over many years (like with digitized news-
papers and archives), it is interesting to
understand how topics have evolved over
time. How did a political situation change? How
did a company start and evolve?
Our demo shows an automatic visualization of
the evolution of a topic over time. We take a
different approach on traditional search result
ranking methods as for example used by Goog-
le. We show a summary over time of the topic searched for. This summary can be used to quickly
get an overview of the back-story of an entity in the text.
How can we visualize the historical background of an entity appearing in a text in a concise way,
without overloading the user with information? This is a hard problem, because of the need to sum-
marize and visualize a large search result without having to cut off the information at a certain point
(such as after the first page of Google search results).
Application
Our main application lies in the visualization of the back-story of entities in news articles. Existing
alternatives include word clouds, which ignore the dimension of time, or stream plots which focus
on the trend comparison of a predefined set of categories, which ignores the open character of
topics.
15. Visualization of topic
evolution in news articles
Jesper Hoeksema
j.e.hoeksema@vu.nl
COMMIT/ project
METIS Dependable Cooperative Systems for Public Safety
Visualization of critical events is applicable in many situations. Our system will also be used to as-
sist operators of the Metis System for Situation Understanding. This system will assist operators in
maritime situations by highlighting vessels that the system deems worthy of attention. The operator
then can investigate the back-story of a vessel using the methods presented above.
Applying the demo to alternative domains would be as simple as finding and converting a suitable
corpus of documents.
Allow a user to see in the definition of a topic over time at a glance.
The definition of a topic or an entity is important, especially how it changes over time.
We provide a visualization of such a change at a glance.
How did a political situation change overtime?
Analyze large sets of textual information by leveraging the ability of the human brain to
spot exceptions to patterns.

I
n the last decade marine traffic has grown
greatly. At the same time terrorist and
criminal activities have moved to the sea
as well. The nations responsible for man-
aging maritime areas therefore more and
more face the problem of detecting suspicious
vessels. Our demo shows how a human op-
erator responsible for managing large complex
maritime areas can be assisted by automatic
risk assessment.
Our system takes inputs from reliable monitoring systems and combines this with in-
formation sources of unknown trustability (such as open source intelligence, pub-
lic databases, websites and social media feeds). The system automatically
presents suspicious vessels to the human operator. This automatic risk assessment can give
a productivity boost to the coast guard, the police, the navy, the customs and to environmental
protection agencies.
How can systems collect and use information to develop a human-like understanding of complex
dynamic situations? More specifically: How can systems automatically determine the intent of an
action?
Current so-called ‘situational awareness systems’ are characterized by cleverly visualized and con-
figurable operational pictures supporting the decisions of the human operator. If the human is taken
out of the loop, the system becomes entirely ineffective. The unique approach of our demo is the
integration of multiple artificial intelligence technologies into a seamlessly automatic operating solu-
tion providing the highest end-user value.
16. Automatic risk assessment of
vessels in maritime areas
Application
The main application is the monitoring of shipping activity and the prediction of the intent of ship-
ping activities. How can we understand the intent of a vessel in a maritime area? The results can
be used by the Thales company in decision support systems for the management of large maritime
areas.
Current systems are fragmented, rely on experience of operators and do not scale well. Continuing
cost pressures are driving the need for intelligent information and reasoning technology. Typical
users can be the coast guard, the navy, the police, customs and environmental protection agencies.
Competitors include for example: 3i:Mind, Greenline Systems and Saab Raytheon.
In addition to safety and security applications our technology is equally suited to providing the
infrastructure needed for Condition Based Maintenance (CBM) systems. Such systems have many
sensors or sources of information which can be used in combination with probabilistic reasoning
models to predict the likelihood of essential maintenance being required in the future. Potential
users are companies designing complex systems, for example ASML, Océ and Rijkswaterstaat.
The hardest challenge is capturing the required domain knowledge.
Safety & security operators will be able to cope efficient with an overload of mission
critical disparate data .. do more, effective, for less.
How to integrate various novel AI technologies into an effective innovative decision
support system.
Innovative technologies for context-aware system architectures dealing with domains
characterized by a certain level of uncertainty.
Using integrated high TRL level demonstrators as means to convince industry of the
value and opportunities of my scientific innovation.
Dave Watts
dave.watts@tno.nl
Bas Huijbrechts
bas.huijbrechts@tno.nl
www.esi.nl/research/
COMMIT/ project

M
ore and more decisions that traditionally
were taken by humans, are taken by in-
telligent machines that perform complex
reasoning. An automatic pilot can fly and
even land an airplane. A medical expert
system can propose diagnoses and treatments
of patients based on their symptoms.
In current systems, only the results of the au-
tomatic reasoning are shown to the user. This
makes it hard for the user to understand and trust the results.
Our demo shows how a user of a safety and security system uses visualization to understand why
and how an automatic reasoning system has reached its conclusions. This allows the user to gain
both a better understanding of the situation and to improve trust in the reasoning system.
How can we best visualize the reasons why an automatic reasoning system has come up with
certain conclusions? The combination of the use of probabilistic reasoning and subsequent visu-
alization of the reasons behind the decision is new and especially important in costly and safety
critical situations.
Application
Our current application area is the safety and security domain. No known products offer visualiza-
tion of the reasons behind the decisions.
17. Let an intelligent machine
explain its decisions
Any system employing automatic reasoning is a potential application. This is especially relevant
in systems where automatic reasoning takes high impact decisions. For example: computer aided
diagnosis systems are used to control the state of an aircraft. If the automated system indicates
that an engine should be replaced, then a human would be expected to be able to carefully check
the reasoning behind such a costly and safety critical decision.
Before one trusts complex high-tech critical decision support systems, they need to
visualize their line-of-reasoning.
How to visualize the decision and reasoning rationale of complex decision support
systems.
New insights in visualizing the internal line-of-machine-reasoning for complex decision
support systems.
Visualization of the inner line-of-reasoning of a state-of-the-art reasoning technique.
Roeland Scheepens
r.j.scheepens@tue.nl
Bas Huijbrechts
bas.huijbrechts@tno.nl
www.esi.nl/research/
COMMIT/ project

I
ndustrial software must continuously be
maintained to stay in business. On the
one hand this is caused by external con-
straints such as changes in hardware, the
advice not to use the outdated Windows
XP operating system anymore, or the Millen-
nium Problem. On the other hand this is caused
by technology changes within the company.
After several years of maintenance, a software redesign is needed to support further innovation.
This is often postponed, because it is time consuming and it does not immediately create new
product features.
Traditional software development is focused on documents, which quickly get outdated. Modern
software development is focused on models that are used for code generation. We develop a
facelift for old software, called ‘model-based migration’. Model-based migration extracts models
from the legacy software and transforms these models to a new model-based design.
The challenge is to obtain a cost effective solution for the huge industrial problem of maintaining
legacy software. Starting from scratch is not acceptable; the domain knowledge must be main-
tained. As documentation is usually outdated, the legacy software itself becomes the primary in-
formation source. This information can be revealed by analyzing the source code and by learning
from the observable behaviour.
Application
We have applied model-based migration to the field service procedures for the interventional X-ray
machines of Philips HealthCare. The legacy software has evolved over many years into a stack of
in-house frameworks, that are difficult to maintain and extend. Our own experience indicates that
an eighty percent productivity gain can be obtained using model-based migration.
The company SemanticDesigns also works on migration of legacy software, but their focus is on
migrations from old programming languages to modern ones. In the applications we look at, the
migration requires a more fundamental redesign.
18. A facelift for old software
Arjan J. Mooij
Arjan.mooij@philips.com
Gernot Eggen
gernot.eggen@philips.com
www.esi.nl/
COMMIT/ project
ALLEGIO Composable Embedded Systems for Healthcare
Intelligent System Design 42 43
In general, all companies that develop software for more than ten years have the same problem.
More specifically, this work has led to the initiative for a follow-up project proposal on software
rejuvenation in the European Ecsel program. Industrial partners from various countries have con-
firmed that they have similar problems with legacy software, and have expressed their interest to
participate in such a project. This project also involves academic partners to further advance the
technology.
Nice to know
The amount of embedded software in high-tech systems is continuing to increase, currently being
in the order of multiple millions of Lines of Code. As a rule of thumb, every fifty thousand Lines of
Code requires one full-time software developer for maintenance only.
Boost innovation by renewing crucial software components.
Reuse existing domain knowledge without suffering from large and complex legacy
software.
Large effort reduction for the maintenance and migration of legacy software.
Challenge is to automate the extraction of domain knowledge from legacy software as
much as possible.

T
hese days, large amounts of data are col-
lected about the operation of many im-
portant systems, for instance, traffic sys-
tems and the financial system. Extracting
meaningful information is very challeng-
ing: big data must be processed in time and
without error.
At TU Delft, for the last four years, we have been collecting data about BitTorrent, a system used
by hundreds of millions of people worldwide for sharing videos and other files. For example, musi-
cians use it for the distribution of their work and software developers for the distribution of open
source software.
Looking at the collected data from all the BitTorrent servers in the world, we can understand
BitTorrent. To do so, we have created a workflow of big data queries that give answers to such
questions as “How many videos are shared?” or “What is the location of the most used servers?” As
our key innovation, we have designed an efficient iterative method to optimize big data workflows.
Despite a large number of empirical and theoretical studies, observing the state of the global infor-
mation networks remains a grand challenge. The main question we set out to answer was how to
reliably analyze large scale time based datasets through different types of queries.
Other questions we address are: What are the programming models to be used? Only the
MapReduce programming model or more? To what dataset sizes we can push our analysis?
Whereas in 2013 we could only process 100 GB, we can now process 1.5 TB, and we are work-
ing on processing all of the 15 TB we collected. With that size, we are among the largest publicly
reported Big Data experiments.
Application
We apply our network analysis to the global BitTorrent network. We have a unique 15 TB dataset
19. BTWorld: A Large-scale Experi-
ment in Time-Based Analytics
Alexandru Iosup
A.Iosup@tudelft.nl
Mihai Capotă
mihai@mihaic.ro
www.pds.ewi.tudelft.nl, www.btworld.nl.
COMMIT/ project
IV-e e-Infrastructure Virtualization for e-Science Applications
obtained from monitoring BitTorrent, which we analyze in one of the largest publicly reported Big
Data experiments. This leads to both the design of a workflow of queries with wide applicability,
and an understanding of the monitored system. The dataset has been obtained by us by periodi-
cally contacting many so called BitTorrent servers worldwide over a period of more than four years.
Our design and experimentation with the BTWorld workflow is applicable to many other time-based
datasets obtained from monitoring large scale distributed systems, e.g., financial systems, traffic
systems and sensor networks.
Quote
“This is a great example of a well-researched and well-engineered real use case of Big Data pro-
cessing.” - Douglas Thain, chairman of the jury for the SCALE Challenge of the IEEE/ACM CCGrid
2014 conference in May 2014. We won this Challenge with BTWorld.
Productivity is increasingly associated with big data. Our innovation allows companies
to pursue ambitious big data projects with complex workflows. E-governance process-
es may benefit to the same degree.
We enable SMEs and research labs with little technical expertise to process big data
in innovative and creative workflows, helping to overcome the data deluge they face.
By 2020, companies will have access to over 40 ZB of data per year. It constitutes as
much a business opportunity as it is a technical challenge. Our TRL 4 technology offers
a critical advantage.
How to enable innovative but non-IT research labs and SMEs to process data with
large volume, high velocity, and significant variety? We offer an efficient, iterative, flex-
ible method.

W
e have developed the crowd emotion
monitor SWEET (short for Sense &
Tweet) that evaluates the emotion of a
group of people based on photos.
SWEET automatically analyzes the emo-
tional expressions on people’s faces as they
appear on images. The images can come from
smartphone cameras or from cameras installed
on fixed locations. Location tracking is done by
using WiFi-sniffers to track the participants. Via
a smartphone app the platform can share the
resulted group emotion with stakeholders.
Our first objective is to enhance people’s experiences during conferences. As a visitor of a confer-
ence SWEET can allow you to see the emotional states in various rooms. You can use this informa-
tion to find out where the most interesting presentation is going on.
How can we reliable track the location of people’s smart phones in a crowd based on WiFi-sniffers?
How can we automatically recognize emotional expressions on people’s faces from images that
come from smartphones or fixed cameras? How should we store and compute the data so to moni-
tor group emotions in real time?
Our emotion monitor combines state-of-the-art sensing, event processing, reasoning and data
communication technologies.
Application
Crowd emotion monitor SWEET can be used at events like conferences to enhance people’s expe-
riences or for entertainment, like at dance parties. More generally, it can be used to guide people
to locations where they can gain the best experience.
There are two ways to use SWEET in practice. First, the SWEET-smartphone app allows you to
make photos of yourself and your fellow participants. The app analyzes facial expressions and
20. Monitoring group emotions
Ben Loke
b.loke@noldus.nl
www.commit-nl.nl/news/commitsweets-bij-editienl
records the location and the surrounding sound. Subjective feedback is retrieved using an event-
driven questionnaire tool about how the participants are feeling.
The second way is to sign up to the special SWEET Event Twitter Channel to receive messages
related to the event. Tweets are broadcasted to inform people about special activities or wherever
something nice or interesting is going on.
Alternatively, the crowd emotion monitor can also be used to increase safety at mass events. By
monitoring the emotion of a crowd and seeing whether the group emotion gets heated up, possible
incidents might be detected in an early stage. However, this alternative application requires the
development of slightly different algorithms. Still another application is the use of SWEET as a tool
to gain insight into group processes, which could be useful in various scientific research.
Nice to know
The first version of SWEET has been demonstrated by SWELL partner Noldus on Dutch television
(Editie NL, RTL4, 16 december 2013)
Local sensing and tweeting as an example of the internet of things.
Crowd Emotion Monitor to assess behaviour and emotion of people.
Use your Smartphone as platform for emotion detection.
Sense emotions locally, tweet worldwide.
Intelligent System DesignCROWD EMOTION MONITOR //
et) will use sensing, event proces-
ta communication to develop a
— an application that is able to
nd emotion of groups of people and
n a highly intuitive manner.
SWEET will enable increased safety at mass events
and enhance the experience of participants in events
such as conferences.
form will be a smartphone using its built-in sensors.
hones will make it possible to measure the behavior
T app will measure:
t’ of Wi-Fi access points to track the participants
eter
quency of the ambient sound
martphone owner or others
ions detected with camera and analyzed using
lysis server
k
T >> GOAL >>
>>
The SWEET project merges technologies from five different COMMIT/ projects: IUALL, SWELL, SenSafety, Metis and TimeTrails.

A
lthough the Digital Age has made ver-
bal communication over long distances
extremely easy, non-verbal communi-
cation (like feelings as expressed by
body language) has received much
less attention. We develop new ways of non-
verbal communication over long distances.
Our demo consist of two ‘whiskers’, placed
separate from each other. The demo looks like
giant whiskers for humans that you can put in
your home and that are connected via the inter-
net to a whisker in the home of a beloved one.
By touching and moving a whisker in your place, the whisker in the home of your friend or relative
starts to move in the same way. This is an innovative way to communicate your feelings in a subtle
way without being physically together.
What is a suitable medium for non-verbal communication using the internet? What is technically
possible? What and how do people want to communicate non-verbally over long distances?
Some existing products and concepts explore this idea, but generally use different modalities for
the input and the output signal. For example, an input signal is translated into a blinking light as an
output signal. Instead, in our prototype the input signal (giving a swing to one of the whiskers) and
the output signal (the other whisker is swinging) are the same. This improves the feeling of being
together.
21. Whiskers let your beloved
ones feel your mood
Application
We develop Whiskers in the first place for elderly people who go through a transition in life. When
an elderly retires, looses a dear one or suddenly has to cope with a physical disability, he or she
might start to feel socially isolated. Whiskers create a new way of feeling socially connected.
As social connectedness benefits everybody, we want to create solutions for other target groups as
well. In that way we also hope to avoid creating a stigmatizing device (“a device only for elderly”),
but a pleasant experience that anyone wishes to have.
Nice to know
Whiskers or vibrissae are specialized for tactile sensing (other types of hair operate as more crude
tactile sensors). Vibrissae grow in various places on most mammals, including all primates except
humans! (Wikipedia).
Quote
“What a good idea to communicate through movements.”
“Whiskers could be used to let others know all is fine without having to call them.”
Loneliness increases the costs of healthcare. One step in solving this hidden problem
is to increase social connectedness between people.
Movement as a mean to feel each other at a distance is new, and could help people to
feel more connected to each other.
Loneliness is a huge social problem. Whiskers are a simple idea to make people feel
more connected when they are separated from one another geographically.
Whiskers support a feeling of connectedness over a distance, without being explicit. It
is like a virtual pat on the head.
Sabine Wildevuur
sabine@waag.org
waag.org/nl/project/commit
waag.org/nl/blog/user-research-de-trein
www.utwente.nl/ctit/research/research_projects/national/COMMIT/COMMIT/
COMMIT/ project
VIEWW Virtual worlds for well-being
Esther van der Bijl
Esther@waag.org

M
any ICT-projects suffer from long delays
because of an unpredictably long test
and integration phase. The problem is
that during this phase many bugs in the
computer code have to be corrected.
Some bugs might lead to a costly redesign.
To detect faults much earlier during the devel-
opment process, we propose a modelling approach based on the Parallel Object-Oriented Speci-
fication Language (POOSL). The new POOSL interface supports light-weight modelling and fast
simulation. In addition, the new interface contains validation checks to detect modelling faults early.
New designs can be validated by simulating POOSL-models in combination with a visualization of
the user-perceived behaviour.
The main challenge is to develop a system modelling environment which allows for fast and light-
weight modelling early in the development process. In this early stage requirements might not yet
be very clear and many decisions have to be taken about the structure of the system, the respon-
sibilities and behaviour of the components, and their interaction.
The modelling environment should also allow industrially scalable simulations. On the other hand,
the desire for model validation and analysis requires a proper formal semantics, i.e. a precise lan-
guage to describe the model. Finally, support for model debugging and visualization of simulations
results is required.
Application
We have applied our method to parts of an interventional X-ray system of Philips Healthcare. The
current way of working is rather document driven. Our proposed new method makes it possible to
simulate and visualize the requirements and high-level designs very early in the development phase.
The modelling tool that will be developed fills a gap between expensive commercial modelling tools
22. Early validation of
software designs
Jozef Hooman
jozef.hooman@esi.nl
poosl.esi.nl
COMMIT/ project
(like Matlab and Rhapsody) that require detailed modelling, often close to the level of code, and
drawing tools (such as Visio and UML drawing tools) that do not allow simulation.
Our approach can be used in all sectors of the high-tech industry, especially for systems where
high-level supervisory control is important. Examples are the command and control systems of
Thales and the lithography machines of ASML.
POOSL has been used already in cooperation with Thales, where performance aspects have been
analyzed. At ASML, engineers are experimenting with a new interface of our modelling tool.
Nice to know
If a problem in the requirements is found only after the product release, it would cost ten to hundred
times more to fix than if it had already been found during the design phase.
Quote
“Modelling the power control protocol of an interventional X-ray system in POOSL revealed a few
design errors that could now be corrected early in the development process. Since POOSL models
can be easily changed, they are very useful to explore the design space quickly. The simulation
possibilities of the POOSL-tool also support the communication with internal stakeholders.”
Increase the rate of innovation of the Dutch high-tech industry.
Fast product development by a shorter test and integration phase.
Avoid costly redesigns by detecting faults much earlier in the development process.
Develop industrially scalable modelling techniques that are maintainable and semanti-
cally consistent.

I
n emergency situations at crowded festi-
vals and parades the 3G- and WiFi-net-
works often fail or become inefficient due
to overload. We present a set of new and
unique wireless sensor network technolo-
gies that help to solve this problem.
We demonstrate an efficient feature detec-
tion algorithm that is, for example, capable of
learning the faces of suspects-on-the-run. We
also show how to build a safe, privacy aware,
emergency ad-hoc network by using the mobile
phones of the visitors.
We show how unmanned aerial vehicles (UAVs) are used at parades or festivals to quickly deploy
the infrastructure of an emergency communication network. Our final goal is to maintain commu-
nication when the standard infrastructure fails.
How can we develop efficient algorithms for detecting important image features? How can we use
the mobile phones of the crowd’s participants to cope with the low bandwidth of current networks?
How can UAVs be used to quickly deploy a network infrastructure for monitoring crowds?
Application
We provide safe, reliable, and privacy-aware wireless sensor networking for crowd and city envi-
ronment monitoring. Our technologies help festival organizers, crowd managers, urban environ-
ment monitoring agencies, and first responders (police, safety staff, and fire brigades) to obtain
information in a faster and more reliable way.
23. Emergency communication
technology for crowd safety
Julio Alexandrino de Oliveira Filho
julio.deoliveirafilho@tno.nl
COMMIT/ project
SENSAFETY Sensor Networks for Public Safety
Information networks are everywhere. Our technologies are also applicable for environmental haz-
ards, military operations, and hard-to-deploy sensor networks. Also the monitoring of urban envi-
ronmental conditions – especially in emergency situations such as explosions or contamination
– can profit from a quickly deployable wireless sensor network. These features make us a potential
partner for military units and environmental agencies.
Nice to know
At festivals or parades, bad news – such as a fire or a fight - reaches Twitter or Facebook earlier
than they reach first responders like police or fire brigade.
Better surveillance of public spaces and crowded events by autonomous drones and
smart infrastructure.
Use drones and smart connectivity to set up or repair your large scale sensor network
in minutes!
Drones and smart phones help bringing emergency news faster to first responders.
Improve your sensor network with flying base stations and smart phone connectivity.

technical design and the interface of the lamps is ready but needs testing. Applications are in the
conceptual phase.
The functions of our intelligent lampposts can be generalized to any kind of situation in which there
is a wish to make people aware of their environment. The lamp can for example show the level
of pollution or noise in a city. And a moving window of streetlights might turn blue to show that an
ambulance is a approaching.
Nice to know
Intelligent lampposts not only enhance safety, but also save energy and produce less light pollution.
Quote
“This looks very promising and we should test the lights on our campus.”
The dynamic lighting conditions provided by intelligent lamp posts, give opportunities
to improve safety and traffic control and are a first step towards the intelligent environ-
ment.
Sensing applications provide a flexible and ubiquitous means to influence the behav-
iour of people in outdoor environments.
Dynamic lighting offers a new means to implement strategies for public safety and
traffic control.
Dynamic lighting conditions provide a new means to communicate with people.
I
t is difficult to inform crowds in emergency
situations because communication is hard
and slow, and because people are pan-
icking. Still, the public should be informed.
The intelligent lamppost (iLP) informs au-
thorities and general public in an intuitive way in
case of emergencies. The lamppost can change
its light by color, intensity, pattern and direction.
Thus it can show the place of an emergency to
first responders or direct people to exits.
The intelligent lamppost is part of a larger eco-
system of autonomous emergency detection and communication. Our demo only shows part of this
ecosystem, namely its lighting capabilities. Part of the demo is a model of a festival terrain with ten
fully functional miniature lampposts demonstrating how they coordinate their actions.
The scientific question is in the area of intelligent interfacing: how to apply intelligent lampposts in
an intuitive way, such that people immediately understand what the lamppost is trying to ‘tell’ them.
Some attempts have already been made to apply intelligent lighting (e.g. in Rotterdam and Gronin-
gen), but not as extended or smart as we propose.
Application
At present there are only lamps available that can change intensity, but they are limited in func-
tion. Our intelligent lamppost goes a few steps further. We develop the intelligent lamppost for the
emergency festival scenario, to inform the public during big events. This scenario shows the crucial
features of sensor networks for public safety: not only lighting, but also communication and smart
detection.
Our research partners are the Dutch police, Tendris, Strukton and the University of Twente. The
24. Intelligent lampposts warn
crowds in emergency situations
Marc de Lignie
marc.de.lignie@vtspn.nl
sensafety.nl
COMMIT/ project
Intelligent System Design 54 55Intelligent System Design

W
e develop modelling and testing tools that improve the quality of complex software. Underly-
ing our tools is the so-called ‘model-based’ testing technique. This is a form of testing that is
not ad-hoc, but based on validated scientific models. Thanks to this approach we can fully
automatize the software testing process.
Our tools contribute to the development of better, more reliable and cheaper software. They
check in a short time whether or not a software/hardware system conforms to its specifications.
Furthermore, our tools make the software development process faster, smoother and more trans-
parent. Finally, they reduce the time to market: the time between the conception of the product and
its availability on the market.
Testing complex software touches on many fundamental aspects of computer science: How to deal
with the staggering amount of discrete states in which a computer can find itself while running soft-
ware? How to deal with non-deterministic aspects, with timing of information flows and with data?
How can we make our testing tools scalable to ever larger software systems? How can we improve
the scientific models with which we describe software?
Application
Our tools are used in a concrete product: the Axini TestManager, a tool that very thoroughly and in
a short time models and tests complex systems.
Competitors: SmarTesting, Conformiq.
25. Automatic testing improves
software quality
Jozef Hooman
jozef.hooman@esi.nl
Machiel van der Bijl
vdbijl@axini.com
www.axini.com
COMMIT/ project
Scalability 56 57
Potential customers are all companies or governmental organizations that develop or use complex
software/hardware systems, for example: ProRail, Philips Healthcare, Verkeer en Waterstaat, and
financial companies like Achmea, ABN-AMRO, Rabobank.
Avoid software bugs that have a large impact on society.
Fast and extensive testing, based on early validated models of requirements.
Reduction of manual test effort and after-sales maintenance.
Improvement of model-based testing requires new techniques to deal with large state
spaces and data.
Scalability
t e s t a u t o m a t i s e r i n g

M
any sectors in our modern society are
producing more and more data: science,
medicine, finance, business, transpor-
tation, retail and telecommunication, to
name a few. Visualization is an effective
way to interpret the meaning of these data.
We develop techniques that greatly speed up
the statistical processing of large amounts of
data. We use these techniques to rapidly visu-
alize the statistical results.
Our demo shows how the interest in a Wikipe-
dia page changes over time. A user can select any set of pages to compare, and also find pages
with similar interest over time. This can for example be used to judge interest in certain topics from
society in general.
How can we speed up the statistical processing of large amounts of data? What are the best visu-
alization techniques for the statistical analysis of large data sets?
Complex statistics are usually limited by the amount of data, since statistical tools are not built
to handle massive amounts of data. We embed a statistical processor into a high-performance
relational database (MonetDB). This combination is unique, as the translation between the two
systems is minimal and thus one hundred times faster than comparable systems. This system has
the potential to deliver new insight into massive amounts of data.
Application
Our application shows the interest in Wikipedia pages over time and is already available online:
http://wikistats.ins.cwi.nl.
26. Rapidly visualizing Wikipedia
page views
Hannes Mühleisen
Hannes.Muehleisen@cwi.nl
Try your own Wikipedia page view visualizations on: http://wikistats.ins.cwi.nl
Watch a video about our work on: http://vimeo.com/groups/amsterdamdatascience/vi-
deos/100491517
COMMIT/ project
Scalability 58 59
Our database contains almost sixty million measurements from nine hundred thousand Wikipedia
pages between January 2008 and July 2014. The data are taken from the Page view statistics for
Wikimedia projects. (http://dumps.wikimedia.org/other/pagecounts-raw/)
Any company or research institute that analyzes large data sets through statistical processes is a
potential customer. Applications range from health care over retail to government.
Nice to know
According to our statistics, people visit the Wikipedia page about ‘Love’ four times more often than
the one about ‘Money’. There is some hope after all…
Quote
“If you start analyzing big data with R + MonetDB, you will no longer have to wait around long
enough to take a coffee break after running each analysis step.” − Anthony Damico, Statistical
Analyst at the Henry J. Kaiser Family Foundation
We produce statistical analysis results 100 times faster than before.
If the amount of data you have to handle for statistical analyses exceeds what your
tools can handle, and you are already using R, we can help.
Our high-performance combination of R and MonetDB is published as Open Source
software and can be used for your projects without licensing fees.
We study how the combination of statistical analysis tools and analytical data manage-
ment systems can yield highly improved performance.
Scalability

C
ompanies, governments, organizations
and scientists have access to more and
more data. But only a few of them have
access to enough statisticians, enough
visualization experts and enough pro-
cessing power to explore the full richness of all
these data.
To solve this problem, we have developed a
data exploration tool called Blaeu. Our system
is named after the famous 16th century Dutch cartographer Willen Blaeu. Our 21st century Blaeu
is a digital ‘data cartographer’.
Blaeu works as a graphical user interface which guides non-technical users through their data.
Through a combination of cutting-edge database technology, smart machine learning algorithms
and effective visualizations, Blaeu will change the way people explore their data.
How can users who know close to nothing about databases become data scientists?
Application
We have initially developed Blaeu to help astronomers to explore their databases. This domain is
challenging, because we have to deal with huge volumes of data, and because we may not even
know what we are looking for.
27. Lost in your data?
Let Blaeu give you a few tips
Thibault Sellam
thibault.sellam@cwi.nl
https://www.monetdb.org/
COMMIT/ project
Scalability 60 61
Our technology can help anyone with quantitative data. We love science, marketing, finance and
the chemicals behind good wines.
Our technology makes data processing maths and algorithms easy to use. Therefore,
more people can understand and monetize data.
Our system helps people write database queries with machine learning and visualiza-
tions. Instead of writing code, people select and zoom into maps.
Our prototype is an unprecedented attempt to make machine learning available to any-
one with numerical data.
Our demo combines in-memory analytics, subspace clustering and hierarchical visuali-
zations to enable fast database exploration.
Scalability

I
n order to predict the earth’s climate, we need to understand the interaction between the
atmosphere (air) and the oceans (water). Only at a resolution smaller than two kilometres
essential physical phenomena such as ocean eddies are resolved in the ocean models.
We develop ways in which climate modellers can use the enormous computing power that
they need for high-resolution and long-running modelling. As high-resolution climate models
require great computational power, we use Graphics Processing Units (GPUs) to perform the com-
putations.
Our demo consists of two parts. The first part is a short movie that shows the results of our ICT-
research: performance models for overlapping GPU-computations with CPU-GPU communication.
The second part is an interactive visualization of the Parallel Ocean Program, a scientific research
program to model the oceans. Our performance models have been applied in this research pro-
gram.
How to optimize data transfers between hosts and GPUs?
Real programs contain dozens of kernels, i.e. small computer programs that manage input-output
requests. On GPUs, the computational time of these individual kernels can often be optimized and
reduced to virtually zero. At that point the transfer times between all these GPU kernels become the
next bottleneck. The problem is that there are many different mechanisms for these transfers and
the best mechanism depends on details of the algorithm.
To solve this problem, we have developed a generic performance model that greatly helps in decid-
ing which mechanism is optimal, thus avoiding the need to implement and measure all alternatives.
Application
We apply the results of our ICT research in the Parallel Ocean Program (POP), a component of
the coupled Climate Earth System Model, maintained by NCAR (Boulder, CO, USA). Our scientific
partner in The Netherlands is prof. Henk Dijkstra from the University of Utrecht.
28. Predicting the earth’s climate
with Graphics Processing Units
Ben van Werkhoven
ben@cs.vu.nl
www.projects.science.uu.nl/esalsa/
COMMIT/ project
Scalability 62 63
The results of our research are being applied to a number of other domains, in particular in applica-
tions where CPU-GPU data transfers form a significant bottleneck.
One example is the problem of Radio Frequency Interference mitigation in Radio Astronomy. An-
other example is deep analysis of data obtained from crime scenes, in particular image and video
data. In this latter domain of forensic analysis we have set up a collaboration with the Netherlands
Forensics Institute (NFI).
Nice to know
The GPU work was nominated for best paper at IEEE/ACM CCGrid’2014 (May 2014, Chicago,
USA) from a total of 283 submissions. So, it belonged to the top one percent.
The work with Utrecht and several other international partners was winner of the Enlighten Your
Research - Global 2014 Award (November 2013, Denver, CO, USA).
To understand the impact of climate change, researchers need high-resolution simula-
tions, which require so much computational power that we use Graphics Processing
Units to perform these computations.
This project presents a performance model that allows application developers to iden-
tify what implementation strategy to use when integrating Graphics Processing Units
into applications.
The ability to efficiently use Graphics Processing Units will allow us to perform Climate
simulations at extreme resolutions, where essential physical phenomena are fully re-
solved in the models.
Our work on performance models for overlapping CPU-GPU data transfers opens up
several avenues for future work, while our work on climate models will lead to more
insights in future climate.
Scalability

O
ver the decades the number of computers and the data traffic has grown so enormously
that the present Internet faces both problems of collapse and security. To cope with these
problems we investigate a fundamentally new concept to organize the Internet: ‘Internet Fac-
tories’. The idea is that with a push of a button one can manufacture a specialized internet.
Essentially these internets can be used similar to the Internet as we know today. Yet our
research allows groups of companies to manage the generated internet. Their specialist software
gives the generated internets advanced properties.
Our research has resulted in prototype software that continuously defends a generated internet
against cyber criminals. Our research enables an industry that takes care to provide all specialist
networking properties our advanced Internet applications require. We will have more privacy, bet-
ter bank safety and we will drive smart cars and use a smart electricity grid because each of these
applications will use their own optimized internet.
We envision distributed applications that are typically deployed over hundreds if not thousands of
nodes using the services of tens of cloud providers spread having data centers at hundreds loca-
tions around the world. This leads to a number of scientific challenges: How can we manage the
state of large distributed applications? What are the limits of their controllability?
Application
Complex distributed applications are created by collaborating, multi-disciplinary groups. The inter-
working of the distributed application is enabled by well-known Internet technologies. Our contribu-
tion is a set of technologies that casts such a distributed application in a form that makes it manage-
able by control software that is also provided by us. The software controls all domain independent
qualities of the distributed application. We demonstrate this with a distributed application from the
Amsterdam Medical Center (AMC) that analyzes different types of brain scans.
29. Building the Internet
of the future
Marc X. Makkes
m.x.makkes@uva.nl
COMMIT/ project
Scalability 64 65
Our implementation is generic and suitable to create applications that control new, complex ma-
chines, geographical large scale infrastructures and distributed satellite systems. Currently we use
our technology with KLM and Ciena Networks to create applications that can defend themselves
against cyber attacks.
Nice to know
Our view on the Internet of the future is that we should program the Internet as a whole. And pro-
gramming the Internet as a whole is like programming an ant heap instead of individual ants.
Here we show how healthcare research and industry can profit from our ICT develop-
ments, and argue how these technology enhances the overall cyber security.
Our technology enables the creation of robust and secure application on more then a
1000 location over the world.
We demonstrate that our technology enables development of normal web-applications
while independent development, using our software libraries, can optimize and add
robustness to these applications.
Here we show the concept for next generation of Internet applications and engage in
discussing future research internet applications and development.
Scalability

T
elehomecare involves the delivery of
healthcare services to patients at home
through the use of ICT. Based on centrally
collected data through remote patient
monitoring, specialized ICT-devices at the
patient’s home provide healthcare services
such as suggestions on sports activities or
medication. The utility of those suggestions is
determined by the collected highly sensitive
patient data, which causes major concerns
about the privacy of the patients.
Our demo realizes telehomecare services while keeping all patient data confidential at all times.
This increases patient trust into telehomecare and gives healthcare providers the technical means
to comply with modern data privacy laws.
How can we use cryptography to protect user privacy in data-mining tasks? In theory, the crypto-
graphic technique of secure multiparty computation (SMC) allows multiple parties to jointly com-
pute a given function on each party’s input without any party learning the inputs of the other parties,
thereby protecting the privacy of each party. In practice however, this general-purpose approach
often turns out to be inefficient.
For the first time, we have constructed a highly efficient SMC-protocol tailored for the computation
of an ID3-decision tree. This tree can then be securely evaluated to form specific predictions.
Application
An ID3-decision tree is a commonly used data-mining algorithm used to provide medical advice to
a patient based on sensitive data collected from other patients. With our partner, the rehabilitation
30. Telehomecare that protects
people’s privacy
Andreas Peter
a.peter@utwente.nl
wwwhome.ewi.utwente.nl/~petera/
COMMIT/ project
THeCS Trusted Healthcare Services
Scalability 66 67
centre Roessingh, we use our privacy-preserving technology to realize the teletreatment of COPD-
patients in a privacy-friendly way.
Our technology is also broadly applicable for computer vulnerability analysis and mitigation. Poten-
tial partners are small and medium enterprises that lack the financial means for IT-security experts
helping to recover from cyber-attacks. On the basis of collected sensitive data on cyber-attack inci-
dents and their mitigation, our technology can automatically provide advice on how to mitigate cer-
tain attacks without expensive experts. Except for specialized graphical user interfaces, no heavy
technical development is needed to bring this into practice. In the Netherlands, this service could
be offered by the National Cyber Security Centre (NCSC).
Nice to know
To show practical feasibility, our prototype has been tested with real patient data from an automated
telehomecare system of the Dutch rehabilitation center Roessingh.
New technology to keep sensitive data confidential in complex data mining tasks ap-
plied to the setting of telehomecare to protect the privacy of patients.
Mechanisms that reduce the rewards of cyberattacks and surveillance by keeping valu-
able data confidential/encrypted at all times.
Our high-performance combination of R and MonetDB is published as Open Source
software and can be used for your projects without licensing fees.
Novel cryptographic approach to efficiently perform data mining tasks in the encrypted
domain, thereby keeping all underlying data confidential.
Scalability

A
lthough there is a great demand for ICT-
supported health care services, both pa-
tients and professionals are only reluc-
tantly stepping in. A main obstacle is trust.
We develop a web-portal that enables the
delivery of healthcare services while incorporat-
ing and evaluating new trust technologies. Ex-
amples of these trust technologies are: data re-
liability evaluation to support decision-making,
transparency tools to show how data is han-
dled, and data leakage and secure multiparty computation for keeping patient data confidential.
Our trusted healthcare services will be used and evaluated in the daily care setting of the Rehabili-
tation Centre with COPD and post-cancer patients – a living lab setting – so not just an academic
platform.
Which technologies can contribute to greater trust in the use of ICT-supported healthcare services?
Can we demonstrate these technologies in a realistic setting? Can we develop a reliable trust
measure for ICT-supported healthcare services?
Application
Our trusted healthcare services include traditional modules like treatment planning and information,
communication facilities and links to the Electronic Health Record (EPD). But they also incorporate
state-of-the-art treatment modules like web-based exercising and an Personalised Activity Coach.
This ambulant coach uses streaming sensor data and adaptive personalized feedback to achieve
a physically active lifestyle.
The platform is developed in strong collaboration with Roessingh Rehabilitation Centre, Roessingh
Research and Development and with the IT-companies VitalHealth and JC Groep, so it is solidly
grounded in the real health world with stable ICT technology, advanced treatment modules and a
firm scientific basis.
31. Trusted healthcare services
Scalability 68 69
By the way we develop and test the platform, we are creating an excellent starting point for suc-
cessful valorization. We aim first at the Dutch rehabilitation market. A second market involves the
care institutes.
Nice to know
Roessingh Rehabilitation Centre is leading in the use of tele-rehabilitation services in the Nether-
lands and participates in several European projects to develop new services.
Quote
“Based on the gained experiences, we will be able to develop quickly a patient portal for the reha-
bilitation centres in the Netherlands” - JC Groep
Only when e-health services are fully trusted by healthcare organizations and citizens,
they will be massively used, resulting in reduction of health care costs and significant
business opportunities.
We develop trusted health care services in a living lab environment, meaning that new
modules can easily be integrated and tested the next moment by users, providing fast
feedback on your product.
We deliver trusted healthcare services – co-designed by healthcare professionals and
patients – with basic and advanced treatment modules, embedded in a living lab envi-
ronment in daily rehabilitation care.
We offer a flexible platform to investigate new technologies that contribute to greater
trust, and demonstrate and validate these technologies in a realistic living lab setting.
Scalability
Hermie Hermens
h.hermens@rrd.nl
Monique Tabak
m.tabak@rrd.nl
COMMIT/ project

B
iomedical research is facing Big Data challeng-
es. At present however, researchers don’t have
user-friendly IT tools to handle these data. To
solve this problem, Science Gateways are de-
veloped. Science Gateways are built as easy-to-
use, web-based and scalable tools that manage and
integrate data, methods and infrastructure for scien-
tific research.
The AMC Science Gateways for biomedical research
enable scientists to run large-scale data analysis eas-
ily and efficiently from web interfaces. From the gateway interface the scientist can start, monitor
and inspect results of the analysis and collaborate with other researchers. The details of the under-
lying data and computing infrastructures are totally hidden from the researchers, so that they can
focus on their research.
How can Science Gateways deal with the perplexing amount and variety of system components?
How can Science Gateways deal with the fact that the requirements from scientists are unknown or
changing? Our research adopts principles of design science for information systems: we iteratively
build science gateways, validate them in the field and generate methodologies and best practices
for the construction of future gateways. Our approach is unique because our gateways are de-
signed for, evaluated by, and adopted by researchers in daily practice. Furthermore, our methodol-
ogy and technology enables fast construction of new gateways across scientific domains.
Application
Our gateways will support the following domains and communities:
Neuroscience: AMC Brain Imaging Centre utilize our gateways to parallelize execution of computa-
tions for neuroimage data analysis and management.
32. Web-based tools for handling
biomedical Big Data
Silvia D. Olabarriaga
s.d.olabarriaga@amc.uva.nl
www.ebioscience.amc.nl
www.ebioscience.amc.nl/gateways/
Watch our videos on YouTube: https://www.youtube.com/user/ebioscience
COMMIT/ project
Scalability 70 71Scalability
Molecular docking: AMC Medical Biochemistry department uses our gateway to perform in vitro
drug screening, which requires parallel execution of massive computations. DNA sequencing data
management: the Bioinformatics Laboratory and the Department of Rheumatology and Clinical
Immunology of the AMC will utilize our gateway to manage ImmunoGenomics data.
Various other groups have shown interest in Science Gateways because they simplify access to
complex data and compute infrastructures. Also small research groups in other organizations from
academia or industry could benefit from this approach.
Due to our methodology and technology, the customization to other applications is straightforward.
For example, the Protein-docking Gateway was built in only three months by an intern reusing the
Neuroscience Gateway components.
Nice to know
During the past three years, our Science Gateways have been used by almost thirty researchers in
the AMC. Together they computed approximately fifteen CPU-years and processed and generated
three terabytes of data. They published so far five papers about the result of the data analysis that
they have performed via the AMC Science Gateways.
Better, faster, cheaper biomedical research. Our Science Gateways enable research-
ers to handle their biomedical Big Data and harness the power of Big Computers with-
out bothering about the IT-complexity inside.
Our Science Gateways are the new enterprise systems between researchers and Big
Data and Computers. They integrate data, people and IT-infrastructure behind easy-
to-use web and mobile interfaces.
Biomedical research needs tools to exploit Big Data and Computers more effectively.
Our gateways provide web and mobile access to data and high-capacity infrastructure
enhancing biomedical research.
Science Gateways facilitate dealing with Big Data and Computers, but they are difficult
to build. We research methods to construct science gateways by iteratively building
them and validating in the field.

companies of the institute in order to understand their needs. These companies include Unilever,
Danone, Nestlé and DSM.
Data management is in its infancy in food research, limited to particular tasks, such as lab man-
agement systems, or to generic document systems. These systems have difficulty in exchanging
data and don’t allow the entire research workflow to be traced. Support for integrating datasets is
non-existent.
Our work is easily extendable to research in other domains, both academic and industrial. Partners
could be other research institutes and their industrial partners, research funding bodies or a com-
mercial ICT partner to market the developed tools.
Our experience in developing semantic tools for research support is unique, due to the combination
of theoretical research and the development of concrete practical tools. Further development would
be required to improve the robustness and user-friendliness of our tools.
Nice to know
Our tools Tiffany and Rosanne were nominated for the 2012 COMMIT/ Valorisation Awards. This
is a national award given by the Dutch organisations ICT~Office, Stichting ICT~Milieu and the
Centrum Wiskunde & Informatica (CWI).
Tiffany: better food research for the taxpayers money.
Semantic tools for sharing research data will revolutionise how we feed our society.
Taking time to invest in research data management makes the difference between re-
search that looks nice on the shelf and research that changes people’s lives.
Properly organising my research process gets the most out of my work.
E
ffective food research requires that data and
methods are shared. At present, careful data
management is considered as a burden rather
than a tool for good science. As a result, data can
no longer be found or interpreted once time has
passed. Furthermore, potential synergies slip through
the net and costly duplications and mistakes occur.
We have developed two tools for the easy manage-
ment of food research data. The first tool, Tiffany, helps
researchers to document their data in such a way that
others can easily trace, understand and reproduce the
research process. The researchers can use a second tool, Rosanne, to annotate their data in order
to further improve search and reuse. Together, Tiffany and Rosanne increase the chances of suc-
cessful valorisation of food research.
Computers excel in data processing but do not understand the data. For example, they can find a
scientific paper containing some given keywords, but they cannot tell where the conclusions in the
paper came from. They can calculate the formulas in a spreadsheet but they cannot find related
datasets and combine them.
To support the researcher in these tasks, the computer needs machine readable models of the
real world. Challenges include: How can we develop a model that supports information exchange
without restricting researchers? How can we embed this support in a user-friendly manner that fits
in the everyday research practice?
Application
We work together with the TI Food and Nutrition, a public/private partnership for long-term strate-
gic research to enhance innovation in the food industry. We also cooperate with the member food
33. How to get more out of
food research
Mari Wigham mari.wigham@wur.nl
www.wageningenur.nl/nl/Expertises-Dienstverlening/Onderzoeksinstituten/
food-biobased-research/Onderzoeksprogrammas/Gezonde-en-smaakvolle-voeding/
Informatiemanagement.htm
Watch our video on: http://youtu.be/cQIdHTwPL1Y
COMMIT/ project
e-FOOD e-Foodlab
Scalability 72 73Scalability

M
onitoring the movements of crowds in
cities can lead to improved city plan-
ning, more efficient traffic flows and safer
crowd management. As camera surveil-
lance might lead to privacy violations, we
use wireless sensor networks to measure who
is close to whom. In particular, we use ordinary
smartphones as sensors. We make sure that
the data of individuals are anonymized.
Our demo shows a website acting as a dash-
board for live streams from downtown Arnhem, where tens of WiFi-hotspots are tracking smart-
phones as they move. The trajectories of pedestrians are mapped onto the actual city plan of the
downtown area. We don’t know who the pedestrians are, nor can we find out: all smartphone iden-
tification is irreversibly transformed into a single cryptographic number before it enters our tracking
system. Given such a number, it is impossible to say to which smartphone it belongs, let alone that
we can identify its owner.
In our demo we will also show screen recordings of trajectories during the World Living Statues
Festival on 27 September 2014, which is expected to attract some three hundred thousand people
to Arnhem.
How can we reliably detect mobile devices and realistically project their trajectories onto a city
plan? One of the problems is that there are many false and missed detections, originating from
very different sources. Identifying trajectories is difficult as there may be many alternatives paths
between two subsequent detections of the same device at different locations.
34. Wireless crowd monitoring
in Arnhem
Maarten van Steen
steen@cs.vu.nl
COMMIT/ project
EWIDS Very large wireless sensor networks for well-being
Scalability 74 75
Application
Our main application is crowd monitoring in highly populated areas. Our partner Foundation Wire-
less Arnhem provides WiFi-hotspots for detection. Members of Arnhem City will use our application
to obtain insight in crowd movements during the World Living Statues Festival.
Our live streaming of hotspot-based crowd detections is unique. Urban sensing as such is not
unique, but the way we support crowd management has not yet been reported.
Our research can also be applied to identify different groups of people visiting a city, such as shop-
pers, tourists and commuters. It can provide much better insight in the long-term movements in a
city, which can be used for city planning (traffic, parking spaces, bicycle trajectories). It can also be
used to correlate shops and venues: e.g. if Alice goes to A, will she also visit B? Natural partners in
our research are other cities, their shops and venue organizations.
Monitoring crowds is applicable to many things and not limited to humans.
Nice to know
It is expected that in 2017 seventy percent of the world population will own a smartphone. Most of
these will be WiFi-enabled.
We automatically measure the movement of crowds in urban spaces.
Our data will allow for the development of a to-do recommendation app.
With this system you will be able to optimize urban/pedestrian logistics.
We need new real-time and scalable mining techniques to deal with noisy data on
proximity graphs.
Scalability

E
ven in case that a GSM network is down
(like during the Enschede fireworks disas-
ter) or overloaded (like during recent fes-
tivals), it is important for public safety that
smartphones can still be used for emer-
gency text messaging. In this way the authori-
ties can inform the public about the emergency.
We have developed a way to realize this emer-
gency communication.
Our protocol is called COCOON: Context-aware
co-operative opportunistic network. COCOON
can be used on any phone and is primarily being
developed for the emergency festival scenario,
to inform the public during big events. Apart from emergency scenarios, COCOON can also be used
by normal users in a ‘Twitter-style’: to send and receive short messages in a peer-to-peer fashion.
How can smartphones still be used for emergency messaging in the case that the GSM-network
is down? What protocol do we need for this? How can we optimize its performance and the effec-
tiveness? How do we make the protocol scalable from a small to a large network of smartphones?
To solve these problems we use the WiFi (or Bluetooth) radio installed in almost every smartphone.
Although this has been tried before, it could only be accomplished in phones that had been hacked
in order to change the core software of the phone. Such efforts excluded the general public. Our
protocol solves this problem as well.
Application
COCOON can be used on any phone and is being developed to inform the public during emergen-
cies at big events and festivals. The first and foremost partner in our research is the Dutch police.
COCOON is under development and is being tested on a small-scale. We have plans to perform
35. Using smartphones for
emergency messaging
Okan Turkes
o.turkes@utwente.nl
COMMIT/ project
Scalability 76 77
medium- to large-scale tests in 2015. At present, the protocol code is stable, but further testing is
needed for optimization.
There are some similar products available, but they all have limitations: OLSR (only hacked
phones), Bluetooth communication (limited number of participants, needs previous pairing), WiFi
Direct (needs previous pairing).
Our emergency text messaging protocol has many applications besides the emergency scenario,
for example: Twitter-style messaging, advertising by shops or stands, public displays, city promo-
tion and public transport.
We plan to make a public programming interface to COCOON, so it can be embedded in other ap-
plications as well, for example in apps used by festivals. COCOON can then also be used to give
real-time information on any changes in the program, to let festival visitors give feedback and to
give festival visitors the possibility to communicate with others in a Twitter-style.
Nice to know
During Pinkpop it sometimes took more than an hour for an sms to reach its destination.
At Pinkpop 2014 neighbors to the festival were asked to open their WiFi access point for visitors
of the festival.
COCOON provides communication in emergencies when everything else fails.
The COCOON platform gives your app direct and free communication with other smart-
phones.
COCOON is a new way to communicate, available for any modern smartphone.
COCOON is an opportunistic delay-tolerant short message communication protocol for
smartphones.
Scalability

E
vents like concerts, festivals and sport-
ing competitions often attract a crowd of
people. The same can be the case for
institutions like museums, hospitals and
amusement parks. We have developed a
real-time visualization of how the density of a
crowd changes.
In our demo we will ask volunteers from the
audience to wear our electronic badge. This
badge will monitor the surrounding density of
its user in real time. The measured densities
will be made visible on one or more screens.
By giving instructions to the volunteers, we will also be able to show real-time changes in density
measurements. All these measurements are done in the network, so that there is no offline, central
server needed.
The scientific challenge is how to reliably estimate the number of people that are in the neighbour-
hood of each person. Each person is a node in a constantly changing network. This estimation is
a scientifically hard problem, because we consider mobile networks with high densities: each node
has typically hundreds of neighbours. An additional challenge is that all nodes estimate the number
of neighbours at the same time.
Application
Tracking the density of crowds in real time can contribute to a safer and more efficient way to deal
with crowds. Our estimator (called Estreme) can at the same time estimate the neighbourhoods
of one hundred nodes with an error below ten percent. Competing solutions presently provide a
36. Measuring crowd densities
for safety and efficiency
Marco Cattani
M.Cattani@tudelft.nl
cattanimarco.com/category/publications/
COMMIT/ project
Scalability 78 79
similar accuracy, but only on networks consisting of a few tens of nodes, where only a fraction of
them estimates the number of neighbours at the same time.
Due to its simplicity, our algorithm can also be used in wristband devices. This makes it more practi-
cal and attractive for potential customers than using the badges, as in our demo.
Nice to know
We have implemented our crowd density measurement tool in an open source operating system for
the Internet of Things. We will soon use it at the Cobra museum in Amsterdam to monitor the flow
of visitors and identify hot spots.
Events like concerts, festivals and sporting events often attract large crowds. Estreme
provides a simple solution for real-time visualization of crowds, which is important for
planning and safety.
Estreme is a low cost wristband that can be given to all attendees in an event. This
wristband preserves the privacy of people but allows a real time monitoring of the
crowd.
Estreme tracks crowds in real time, which makes an event safer and more pleasant.
It can cope with higher densities than existing solutions (100’s of neighbors) with the
same low error (below 10%).
Estimating the density of crowds is a hard problem because it requires monitoring
mobile networks with high densities. Estreme uses periodic but asynchronous radio
signals to solve this problem.
Scalability
Van Mierlo

I
nteractive playgrounds are installations
that combine the benefits of traditional
playgrounds for children with advances
in technology. In our demo we show the
Interactive Tag Playground (ITP), a mod-
ern, interactive version of the century-old game
called ‘Tagging’ or ‘Tikkertje’ in Dutch.With the
Interactive Tag Playground we actively steer the
interaction between players. To this end, players
are being tracked using Kinects and additional
body-worn sensors. In its most basic version we
project differently colored circles around each
player to indicate the role of tagger or runner.
When these circles collide, a ‘tag’ is detected
and the roles of the players switch. We also add
novel interactive elements such as power-ups
and bonuses. Apart from entertainment, the Interactive Tag Playground is also a tool to study how
children interact with each other and with the environment. Our final aim is to automatically steer
the interactions in such a way that all players remain engaged and physically active.
How can we best track the players and understand their interactive behavior? What kind of interac-
tions are suitable at which moment during the play to maintain the engagement of players? How
can players be motivated to adopt certain play styles such as cooperation or competition?
Application
Our work can eventually be combined into playgrounds that allow for adapted play without any
restrictions such as the need to wear certain sensors. We see great potential for these playgrounds
in open spaces such as traditional playgrounds, shopping malls, sporting facilities and outdoor
squares.
37. Tikkertje 2.0
Robby van Delden
r.w.vandelden@utwente.nl
Alejandro Moreno
a.m.morenocelleri@utwente.nl
COMMIT/ project
Physical User Interaction Styles 80 81
Interactive play is everywhere. Not only children can benefit or enjoy interactive play. Previously, we
made an ambient interactive bar that was successful in entertaining adults. The techniques, both
for sensing and interaction, are largely the same. One can also think about offering play to people
who normally have limited play opportunities, such as heavily handicapped or demented people.
Interactive play might help in socializing, entertaining, triggering physical activity or even branding.
Nice to know
The behavior of adults and children while playing tag is largely the same. Except that children cheat
more.
In 2008, the UK Local Government Association promoted tag games stating that children are over-
protected (“wrapped in cotton wool”).[Wikipedia]
Variants of tag have fancy names such as “Cops and Robbers”, “Kiss Chase” and “Ringolevio”.
Quotes from participating students:
“This looks so cool!! I want to play!”
“I want to have this at home!”
“This should be a new sport, Olympic”
“I think this would fit very well in gyms or fairs”
An interactive whole-body game aimed at enhancing the player’s fun, physical and
social experience.
An interactive game environment that senses players’ behavior to facilitate a physi-
cally active, social and engaging experience with multiple other players without prior
calibration.
A novel game environment that mediates gameplay elements by sensing and tracking
players to promote targeted behavior or improving the game experience.
Tracking and analysis of human behavior to create engaging gameplay.

cy. There is a growing market for running applications, like Endomundo, Runkeeper, MapMyRun,
Zombies-Run and Nike+ Running. Also, sport watch brands like Garmin, Polar, Timex, TomTom,
Suunto and MIO are competitors for this kind of applications. Although many of those applications
give feedback on step frequency, none of the current brands have individualized feedback on indi-
vidualized optimal step frequency.
Similar measurement approaches to ours are currently used in other sports, such as rowing and
speed skating, and in medical applications, such as assessment of movement disorders in neuro-
logical patients and of fall risk in elderly. Our algorithm is innovative both in terms of real time use
and in robustness, and can be of benefit in these fields as well. Similarly, combining accumulating
individual data with group-based reference data is likely to be useful in many sports and medical
applications and possibly in a much wider field.
Nice to know
Because step frequency remains relatively consistent across different running event lengths, it is rec-
ommended to first optimize step frequency and later adjust step length to obtain the desired speed.
We provide feedback of current individual optimal stride frequency by changing (music)
beats to the desired rhythm. In this way, the runner is stimulated to unconsciously run
with an improved technique to prevent injuries.
We combine data sources, like heart rate, phone sensors and data from previous runs
to provide the runner individualized feedback about their optimal stride frequency.
No expensive devices needed, you can run with your smartphone and still get accurate
individualized feedback. Our application has a robust stride detection algorithm and
uses previous runs to change music beats to your current optimal stride frequency.
Heart rate of previous runs is used to find an optimum stride frequency for different
speeds for an individual. Higher stride frequencies lowers impact forces and therefore
is expected to decrease injury. We developed a robust algorithm that detects strides for
a broad range of periodic signals for unconstrained (phone) sensors.
E
very year about one-fifth of the 2,5 million
Dutch amateur runners gets injured. Run-
ning with too large steps, and therefore
mostly with a frequency that is too low, is
known to increase the injury risk of a run-
ner. The optimal step frequency of a runner de-
pends on the heart rate and the running speed,
and differs between individuals. Beginning run-
ners are known to have step frequencies below
their energetic optimum.
Using the smartphone as a sensing tool, a beginning runner can be guided to increase step fre-
quency if necessary. From previous runs an optimal step frequency can be calculated. To this end,
we develop a robust step frequency algorithm for unconstrained smartphones and calculate indi-
vidual optimal step frequencies from training data.
How can we use parameters measured of a variety of signals to optimize a desired response?
Robust algorithms are needed to work beyond well-controlled environments of a laboratory. And
another challenge is the time-variations in the signal.
Application
The application centers around the question how to use parameters as heart rate, step frequency
and speed to optimize the technique of an individual amateur runner? The optimal step frequency
of a runner can change over time, due to training or injury. This requires an optimization process
that combines group-based data with individual data of the runner. This has been done previously
in laboratory research at slow running speeds, but finds its novelty in a broader range of speeds.
As the optimal step frequency of a runner can change over time, due to training or injury. This re-
quires an optimization process that combines group-based data with individual data of the runner.
Individual amateur runners can use our application to help them finding their optimal step frequen-
38. Advising optimal step
frequency for runners
Ben van Oeveren
b.t.van.oeveren@vu.nl
www.move.vu.nl/nl/over-move/medewerkers/O/bt-van-oeveren.asp
COMMIT/ project
SENSEI Sensor based Engagement for Improved Health
Physical User Interaction Styles 82 83Physical User Interaction Styles

The knowledge we gained about the human speech apparatus can also be applied in other areas,
for example speech and language therapy. Some people have difficulty speaking due to breathing
problems caused by diseases such as COPD. Having software programs that can help train people
in their own time to control their speaking apparatus could be beneficial for these people. In addi-
tion, our software can be applied in other smartphone-based speech analysis tasks as well, such
as emotion recognition.
Nice to know
“It’sveryinteresting.Withthespeechdataobtainedyounotonlylearnalotabouthowspeechparameters
changeunderphysicalstress,butyoucanalsodeveloptechnologythatclassifieslevelsofphysicalstress.
Iwouldbeveryinterestedinusingthespeechdatacollectedforclassificationchallenges.”−Anonymous
Assessing your exercise intensity by talking to your phone while running is a fun and
cheap alternative to assessing it by measuring your heart rate through bands or expen-
sive watches
Our voice app offers a fun and accessible way to assessing exercise intensity to help
prevent beginning runners, who often run at a too high intensity, from risking injuries
and program dropouts.
We will not only be innovative in our research, studies on speech under physical stress
are still limited, but we will also utilize the knowledge to advance voice technology to
detect physical states.
Research on changes in people’s speech under the influence of physical activity is
novel. It can be utilized for an app that detects exercise intensity through voice.
W
e have developed a smartphone app that
provides a novel, automated and unobtru-
sive way of assessing a runner’s physi-
cal state based on speech. Beginning
runners often have difficulty determining
whether they are exercising at the right inten-
sity. They often start with running too fast, and
this increases the risk of exercise dropout and
injuries.
One of the most widely used subjective ways to assess the level of exercise intensity is the ‘talk
test’. When you can still speak comfortably while running, you are running at the right intensity. Our
smartphone app analyzes your voice and indicates whether or not you are exercising too hard.
How does speech production change under influence of various conditions? How can we develop
an algorithm that uses this knowledge for reliable automatic voice-based assessment?
To solve these challenges, we train classifiers to gauge two signals while they are being acquired
each under individually varying circumstances.
Application
The app measures the speech production during running exercises. Speech and heart rate data
are collected from people talking while running at various intensity levels. To the best of our knowl-
edge, we are the first to address these two challenges simultaneously.
Through automatic voice analysis, the ‘talk test’ can be made more objective but still in an individu-
alized way. It works unobtrusively as a voice-based app on a smartphone that people often carry
while running. Although there are many running apps available in the market, none of these have
the features as presented in our demo.
39. Run, talk and don’t
get injured
Khiet Truong
k.p.truong@utwente.nl
www.cs.utwente.nl/~truongkp
COMMIT/ project

The algorithm we propose can be applied to other domains as well.
For example, it can be used to help companies around the world to find potential business partners.
If each company would expose its services in a standardized way, our system would be able to
search among hundreds of thousands of companies in order to find meaningful partnerships (e.g.
Apple with Nike). It could also be used to sell bundles of products that are related or to match peo-
ple in social networks.
Nice to know
The National Public Health Institute (RIVM) estimated that, in the Netherlands, 1,4 percent of ex-
penditure on health is a consequence of too little exercise.
It has been estimated that in a big city like Rotterdam, sport benefits for nearly half a billion euro per
year. Slightly more than half is the result of a healthier population, almost a quarter is due to less
absenteeism at work, twelve percent to a greater quality of life and three percent to less school ab-
senteeism.
Our app helps runners find people to run with that have the same interests and fitness,
thus making running more fun and social.
We take an innovative approach to make running more social, by combining the latest
mobile technologies in an app that helps runners make new friends while they run.
Bored of running alone? Our app helps you make new friends that are just as good
runners as you, so you’ll never have to run alone again.
We use a distributed clustering algorithm to group runners in real-time based on their
physical parameters and social interests.
C
urrently, most amateur runners train
alone, which may lead to decreased mo-
tivation over time. We have developed an
app that intelligently groups runners with
similar physical parameters in order to
make the training more efficient and more fun.
In our demo, we display interactively on a
screen how a person running on a treadmill
follows a track with other runners in a virtual
environment. During the run, the app tries to match her with other groups of virtual runners that she
passes by, based on the speed and fitness of the person. When a good group is found, the runners
in the group are coloured in a certain way, to indicate our runner that she should train together with
them.
Can we make an app that clusters signals in groups that are similar? Can we develop a clustering
algorithm that is fast, fault-tolerant and has low battery usage?
As a solution, we propose a novel peer-to-peer clustering algorithm. To our knowledge, this algo-
rithm is the fastest of its kind and the attempt to cluster runners in real-time has never been made.
Application
The context is to monitor runners in real-time based on their physical condition and social interests.
Our clustering algorithm can use different types of criteria for matching, like physical parameters,
interests, age etc. Depending on the chosen criteria, the app can be used to stimulate people to
keep up with the others, thus improving performance or to encourage people to socialize.
There are already other apps on the market that challenge people to run together (e.g. Endomon-
do), but ours is the first one that does this in real-time.
40. Intelligently grouping
amateur runners
Nicolae Vladimir Bozdog
n.v.bozdog@student.vu.nl
www.linkedin.com/pub/vladimir-bozdog/5b/217/b88
COMMIT/ project

the sensation of grounding the body while making different power poses like stretching up. Cylinder
and Kinect camera are programmed to trigger loud noise when ‘the performer’ is not standing in
the desired position on the object. When standing and moving in the right manner, it quiets down.
The Rolling Stairs are a way to explore collaboration between two users. Interaction is focused on
‘balancing’ between different types of collaboration. The Stairs are programmed to trigger single
notes or parts of melodies. Collaboration may lead to melodic lines, rhythms, musical compositions
and finally silence as the ultimate composition. Our balance objects have great potential for use in
special needs education (e.g. related to performance anxiety, ADHD and concentration problems),
mental health care (anxiety and obsessive-compulsive disorder) and children’s rehabilitation (co-
operation, motivation, motor development, neuromotor impairments).
The found knowledge can be applicable in ICT and other domains, like rehabilitation, physical ex-
ercise, personal coaching, presentation, leadership training, etcetera.
Nice to know
Our project was screened in the NTR TV education special ‘10x Beter’: http://waag.org/nl/nieuws/
embodied-learning-op-tv. Our project was nominated for the IPON Award 2014: http://event.ipon.
nl/genomineerden-ipon-awards-2014
Cylinder, Rolling Stairs, and Seesaw are a series of balance objects, developed to
enhance corporal literacy.
We aim at providing a feedback loop that stimulates focus on the moving body itself.
The concept of natural interaction with technology has a great potential for special
needs education, mental health care and rehabilitation.
We promote natural interaction with technology through incorporating senses and the
body.
L
earning in the 21st century puts high val-
ue on cooperation, personal involvement
and creativity on the one hand, and digital
skills on the other hand. In most of these
digital skills the human body hardly plays
a role, apart from typing on a keyboard. How-
ever, it is well known that the body does play an
important role in learning. The mind uses and
grows out of such bodily capacities.
Especially for the use at schools, we have
developed three balance objects (Rolling Stairs, Cylinder and Seesaw) that help teenagers to
learn with their body as an interface. Each object has sensors and is connected to an audio box.
Teenagers can stand on an object and try to reach a balance. Depending on their movements, they
hear different musical sound as a feedback. The sounds motivates them to change their posture.
The objects evoke body awareness and emphasize the natural functionality of our bodies, like
standing up, climbing, stretching and balancing. All objects aim to be silenced by the user, as
silence is taken as the ideal state-of-mind to focus on the embodied self.
How to design interactive devices in which the body itself becomes the interface? What do we learn
about the borders of the body while it interacts with these sensory devices? How easily does the
body consider these objects as part of itself?
Comparing our prototype objects with other devices, using the body as interface like Wii Fit Plus,
the feedback loop that we use is unique. The participant is not only focused outwards, towards the
screen, but the objects constantly lead the focus back to the sensation of the moving body itself.
Application
All our balance objects have a different purpose in an interactive game between the user and the
object. The Cylinder is an empowerment tool focused on finding balance. The participant discovers
41. Learning with your body as
interface
Karien Vermeulen
karien@waag.org
waag.org/nl/project/embodied-learning-installatie
COMMIT/ project
Kennisnet Marloeke van der Vlugt

We have partnered up with Bascule (a psychiatric institute for children and youngsters) to explore
the possibilities to integrate the Moodroom in some of their therapies.
Competitors might be found in developers of Microsoft Kinect applications.
In other therapeutic situations, for example while rehabilitating after a stroke or injury, the Mood-
room could function like a diary to record the recovery process in relation to the emotional health
of the patient.
The installation could also have a place at festivals and other informal places where people can
interact with each other. Visualization can have a more public function. The application can have a
function both in communication and in empowerment.
Quote
‘NICE! I didn’t know that you could show emotions with colours and shapes.’ - Girl (15)
‘The reactions of the youngsters were very positive. Nice to work with emotions in an interactive way.’
- Therapist at the psychiatric institute Bascule
Visual communication of emotions in Moodroom is displaying embodied interaction,
enabling youngsters to express and share their emotions without having to talk about
them.
Moodroom is a physical space where visual communication of emotions can be dis-
played by embodied interaction domains.
Moodroom is a physical space for emotional communication, using a newly designed
‘language’ based on sensory experiences and body movements, which can be applied
in a wide range of domains.
Moodroom is a physical interactive space where wordless and visual communication
of emotions can be displayed, using a newly designed ‘language’ based on sensory
experiences and body movements.
W
e have developed a Moodroom in which
youngsters can express their emotions by
moving their bodies. The Moodroom is an
interactive installation that records bod-
ily movements with a Kinect-sensor and
translates them into colours, patterns, vibrations
and sounds. The visual content is projected on
three screens in front of the user. In the Mood-
room youngsters can express and share their
emotions without having to talk about them.
The Moodroom is based on the principles of embodied cognition. Embodied cognition is an upcom-
ing subject in both social psychology and neuropsychology that recognizes the mutual influence
between our body, senses and cognitive functions. One aspect of this is communicating about
emotions through visual cues, like colors and patterns.
The goal of the Moodroom is to create a new way in which youngsters can get to know themselves
and each other in an embodied learning environment.
How can sensory experiences be translated and used in a playful way in an interactive environ-
ment? First, we focus on abstract visual representations to arrive at a new ‘language’ of movements
in which participants can express and communicate their emotions better than in words.
We have the ambition to add other sensory experiences like sound and music.
Application
Many youngsters experience difficulties in expressing emotions and showing empathy. By using
a new ‘language’ based on sensory experiences and body movements, and with the use of visual
and audio representations, youngsters can express themselves in an accessible and entertaining
way without being hindered by the abstract character of language.
42. Playing with emotions
in the Moodroom
Meia Wippoo
meia@waag.org
waag.org/nl/project/commit
COMMIT/ project
Bascule

down and helps to retrieve memories. It provides a point of reference for people that display
wandering behaviour.
The wall is developed together with care organizations AMSTA and Vivium Naarderheem. It has
been tested at two different locations.
Competing products exist, but these do not use ICT to provide personalized content.
Computer interfacing supported by computer vision is broadly applicable.
The wall could also be used for people who suffer from mental disorders such as severe autism.
The computer vision technology can also be used for human pose recognition and for the classifi-
cation of actions and interactions involving multiple individuals.
The content selection technology could be used to influence the mood of people in public spaces.
Nice to know
Three students of the Hogeschool van Amsterdam (HvA) are exploring business opportunities for
interactive technology for people with dementia through their startup company iLLi-engineering.
Quote
“Ik zou wel willen dat dit geplaatst wordt in het verzorgingstehuis waar mijn moeder woont.”
The Interactive wall improves the quality of life of people suffering from dementia by
reducing wandering behaviour and providing a distraction from depressed feelings.
The interactive wall uses computer vision technology to give elderly suffering from de-
mentia a personalized experience. This helps to retrieve pleasant memories and it
reduces wandering behaviour.
The interactive wall reduces the wandering behaviour of elderly with dementia and
provides a distraction from depressed feelings. And it allows the caregivers to spend
more time on other patients.
Recognize people’s movements and select personalized content to engage people
with one another and with the interactive wall.
P
eople suffering from dementia often feel
confused and depressed. Some of them
also display wandering behaviour.
We build an interactive wall for people suf-
fering from dementia. The wall uses com-
puter vision to recognize the person in front of
the wall and to recognize his or her behaviour
and emotion. Based on the detected behaviour
the wall then gives a personalized experience
using video and music that the elderly like.
Family members can upload content.
The interaction with the wall may diminish the behavioural problems of dementia such as agitation,
aggression, fear, depression and apathy. The wall also gives those who display wandering behav-
iour a virtual place to go to.
How to automatically recognize people, their behaviour and emotions? How to select appropriate
audiovisual content automatically under demanding circumstances?
Application
In the early stages of dementia people are aware of their situation, which leads to depression, while
in later stages of dementia people become passive, they react to some triggers but do not show
initiative by themselves.
The target audience of elderly suffering from dementia presents severe problems. Many of them
cannot explicitly state their own preferences. Furthermore, their facial features and posture may
change rapidly and their behaviour is sometimes unpredictable.
The interactive wall provides distraction from depressed feelings, activates people or calms them
43. Personalized interactive wall
for elderly with dementia
Pascal Wiggers
p.wiggers@hva.nl
www.digitallifecentre.nl/projecten/virtual-worlds-for-well-being
COMMIT/ project

Alternatively, our tactile sleeve can be employed in the entertainment industry. For example, it can
be an additional method of communication in online multi-player games. And while watching a
movie with a distant friend, the smart sleeve can be used as a playful way to enhance the movie
experience.
Nice to know
Do Androids Dream of Electric Sheep? - Early versions of the TaSST used conductive wool (sheep’s
wool with metal fibres) to detect touches.
Quote
“Pak de onderarm van je buurman eens vast. Pets er drie keer stevig op. Vraag dan: welke emotie
probeer ik over te brengen? Wedden dat je buurman zegt: eeeuh, woede?” - NRC Next, 21-07-2012
“[...] de TaSST, een apparaat dat menselijke aanraking simuleert en deze over kan brengen naar een
ander, waar dan ook ter wereld.” - Scientias, 29-12-2012
“Nu communicatie via tekst, geluid en beeld goed op dreef is, is het tijd voor een extra zintuig: tast.”
- Voorpagina artikel, NRC Next, 22-03-2013
The TaSST uses custom designed fabric sensors, that can be reshaped and embedded
in a host of different garments.
Health benefits of social touch are numerous, and mediated social touch could bring
some of these benefits to people who may lack social touch in their treatment.
Mediated social touch is an understudied phenomenon and the TaSST offers research-
ers an open platform to study effects of mediated social touch.
Mediated social touch can play an important part in helping people live independent
lives, with intimate connections to their loved ones, for longer.
W
e do not only use our sense of touch to
explore the world around us, but also in
social interaction. Social touch can help
to reduce stress, communicate feelings
and emotions, and is vitally important for
the bonding between people. Unfortunately,
our everyday digital communication via smart
phones and computers completely lacks the
sense of touch.
To overcome this disadvantage, we have devel-
oped a wearable, smart-textile sleeve, called TaSST (Tactile Sleeve for Social Touch). You can
wear TaSST on your forearm and communicate through touch at a distance. The sleeve can regis-
ter a gentle caress or a strong squeeze, and reproduce these touches as a vibrotactile pattern on
a smart sleeve worn by somebody else. Our tactile sleeve helps to supports physical closeness
between people separated by distance.
How can we use wearable haptic technology to communicate the important sense of touch?
Application
We have designed our tactile sleeve as a research tool to study the communication of social touch.
People separated by distance, such as family members working abroad, patients isolated in hos-
pital wards, and people living in elderly homes can use the sleeve as an additional communication
channel with loved ones. Deaf-blind people can benefit from it by allowing them to more easily com-
municate with their care-givers, and opening up their world to new experiences.
Tactile is one of the six human cognitive sensors.
44. Send your loved one a gentle
caress via a tactile sleeve
Gijs Huisman
G.Huisman-1@utwente.nl
Aduen Darriba Frederiks
a.darriba.frederiks@hva.nl
COMMIT/ project
Bartimeus Elitac

faster or slower step. These changes in walking patterns are required in situations like passing a
doorstep or avoiding a puddle of water on the floor.
Nice to know
Revalidation products based on extensive scientific research are becoming available for homecare,
just in time for the move towards more practicing at home.
The system aims at accelerated rehabilitation by letting people do exercises at home
as part of their daily routine.
The approach measures how well someone varies their walking pattern in response to
environmental changes.
Sensors measure the degree of decreased adaptive capacity. It may predict the likeli-
ness of a higher risk of falling.
How to measure walking patterns in response to the so-called deteriorating adaptive
capacity of a person by use of sensor techniques.
I
n many patients with neurological or ortho-
paedic disorders, but also in many healthy
elderly, the ability to suddenly change the
walking pattern is restricted. These people
have a higher risk of falling, for example
when trying to step over an obstacle. To reduce
the falling risk, we have developed a fun and
motivating exercise game.
During an assessment on a walking belt we
measure the adaptivity of a patient’s walking style. After defining a comfortable walking speed,
we present a visual feedback of a target step length. The subject then needs to respond by taking
smaller and larger steps respectively. The belt speed adjusts automatically to ensure a constant
step frequency. We stimulate the patient to change his step frequency by controlled belt speed
changes in combination with target step-lengths.
Our assessment measures how well the patient performs adjustments in step lengths and step
frequencies. The better the adjustment, the lower the risk of falling.
How can we improve the adaptability to patterns in personal data?
Application
Specifically, how can we reduce the risk of falling by data analysis?
Our fun and motivating exercise game helps to reduce the risk of falling. Patients are challenged to
change their walking pattern constantly in response to their environment. In this way they will train
their adaptive capacity and reduce the risk of falling.
To prevent falls in everyday life, people need to be able to adapt their walking pattern if necessary.
We can change our walking pattern for example by taking a smaller or larger step or by taking a
45. An exercise game to reduce
the risk of falling
Pieter Goossen
pieter.goossen@cgi.com
www.motekmedical.com
COMMIT/ project

or offer only very limited interactivity. Using our implemented techniques, this training can be made
more effective, because trainees can directly see the results of their actions and learn why the
interaction played out as it did.
Serious gaming has a broad impact. Our approach can be used for similar serious games for other
professions that would benefit from social skills training, such as social workers and health care
professionals. Using our approach to serious games, the training programs for these professions
can be made more effective. The technical framework underlying our game enables easy editing
and creation of both scenarios and characters.
Other commercial partners could be companies specialized in assessment and human resource
development.
Nice to know
Currently, only a fraction of police trainees can practice their social skills with professional actors.
Most of their knowledge is gained by accompanying more experienced officers.
The loitering juveniles in our serious game lend their names from the Dutch tv series “New Kids”.
This serious game lets police officers improve their social awareness through training
with intelligent virtual characters so that they are better at handling conflicts in a de-
escalating way
This serious game uses a virtual reality setup to assist in social skills training by letting
trainees gain first-hand experience in how their choices may affect others.
This serious game combines social and computer science to improve people’s social
awareness and paves the way for the creation of other scenarios for social skills train-
ing.
This serious game uses artificial intelligence informed by theories from social psychol-
ogy to guide the behaviour of virtual characters so that players can improve their social
awareness.
W
e have developed a serious game called
LOITER to train the social skills of police
officers. Players of the game have to re-
solve a conflict with a group of loitering
juveniles. Through playing this game,
police trainees can improve their social aware-
ness.
Players interact with virtual juveniles in a 3D-
environment using a full-body immersive virtual
reality system. The virtual juveniles use artificial
intelligence to respond to the player according
to theories from social psychology. Thus, the
choices of the players in how to reason with the juveniles determine the outcome of the conflict.
One of our main challenges is to model human behaviour and let virtual characters respond believ-
ably to human users. To create such a model, we analyzed interactions between police officers
and civilians. We also implemented an interactive story structure that lets people experience the
influence of their own behaviour on others. This results in different scenarios with different story
endings.
Our approach is unique because of its combination of methodologies and the ability of the charac-
ters to explain the reasoning behind their actions.
Application
We have developed LOITER in cooperation with the research centre for serious gaming T-Xchange,
software developer re-lion, and the former national police services agency KLPD.
Existing products that focus on training of social skills for police officers are practically non-existent
46. A serious game for training
social skills of police officers
Jeroen Linssen
j.m.linssen@utwente.nl
hmi.ewi.utwente.nl/IUALL/demos/loiter/
COMMIT/ project
IUALL Interaction for universal access

tween individuals each with different cognitive, affective and social barriers to learning. Practicing in
a virtual learning environment will provide users with requisite knowledge and skills, and increase
the self-efficacy required to participate.
Currently, our vision is to distribute this system for instance in public libraries and classrooms, so
users can use the system regardless of whether or not they own a personal computer.
Our virtual environment technology can also be used in many other application fields: language and
culture learning, distance learning, learning on the job, and entertainment.
New tools can be included that could allow users to create their own content for use in this system.
This would give the system the additional role of a lesson creation support tool for teachers, and a
method of expression for students.
Quote
“Het ontwikkelen van deze virtuele leeromgeving is volgens mij tamelijk origineel en innovatief.” –
Wim Matthijsse, Stichting ABC
Virtual neighbourhood, software that learns how to help people of low literacy and non-
native citizens to better take part in local daily life.
Virtual neighbourhood software modelling the social and physical environment to sup-
port people of low literacy and non-native citizens in engaging with public and private
services in their neighbourhood.
A learning environment software package that employs state-of-the-art didactics, inter-
action and personalization to provide a lifelike virtual neighbourhood for social partici-
pation.
The system provides cognitive, affective, and social learning support using compre-
hensive user modelling and individualization to teach skills and to improve self-efficacy
and motivation.
W
e have designed a virtual learning envi-
ronment that supports low-literates and
immigrants with language problems to
participate in the Dutch society.
In this virtual environment, users can se-
lect prototypical situations in which they need
information and communication skills in both
formal and informal social settings. We have
developed four scenarios: online banking, gro-
cery shopping, visiting a service desk and chat-
ting at a bus stop.
The virtual environment uses virtual humanlike characters that graphically appear on a screen. The
characters engage in a conversation, show gestures and have facial expressions like real humans.
They either conduct conversations with the user or support him in doing participation exercises.
What is the best way to design an effective virtual learning environment? In particular, the learning
environment should be able to deal with large differences in the input.
The virtual learning environment should also adapt the offered content and the difficulty level over
time.
Application
The primary application of this system is in supporting the societal participation learning of low-
literates and immigrants in the Netherlands. Therefore, the system should be able to support each
individual user achieving their intended learning objectives taking into account large differences be-
47. Virtual learning helps
people with language problems
in societal participation
Dylan G.M. Schouten
D.G.M.Schouten@tudelft.nl
commit-nl.nl/projects/wp-packages/social-conventions-learning-in-mixed-reality
COMMIT/ project

G
ood product design contributes to good
business. Unfortunately, many people
have difficulties with using everyday tech-
nological products. To solve this problem,
the end-user should be more involved in
the design process.
We have developed the Include Toolbox to sup-
port small and medium ICT-enterprises in ap-
plying ‘inclusive design methods’ in their prod-
uct development. By involving the end-user in
the design process, inclusive design tries to
include as many users as possible.
The user of the Include Toolbox first chooses a method tailored to the specific design phase and
envisioned user group. Next, the user receives step-by-step support from preparation and execu-
tion of the method to analysis of the results. Several toolbox methods have already been applied,
which resulted in two apps. The first app supports children around the neighbourhood, for example,
by keeping them safe and helping them make new friends. The second app supports elderly to
organize adhoc help when needed.
How can we bridge the gap between theory and practice of inclusive design? The tool needs to be
sufficiently detailed to address all end-user groups in a meaningful way. At the same time the tool
needs to be comprehensible and accessible for both experienced and inexperienced users.
Application
The tool is specifically aimed at small and medium ICT-enterprises. The Include Toolbox enables
the application of scientific knowledge to business practice.
The Include Toolbox is a web-based toolbox where small and medium ICT-enterprises can find
48. Involving the end-user
in the design process
Human Computation 102 103
guidance in involving end-users of their products throughout the entire development process, from
recruiting participants to analyzing the results.
The toolbox does not educate bad designers to become good designers. It supports the small and
medium enterprises with early discovery of possible product failures or exclusion of the user group
of the developed ICT product or service.
Although a few digital inclusive design toolboxes already exist for research and designers, none of
them is adapted to small and medium enterprises that develop ICT products and services.
Design is everywhere.
The toolbox can be extended with other end-user groups and methods, evaluation of methods, best
practices, design guidelines and design patterns.
Nice to know
The Include Toolbox itself is designed by including small and medium ICT-enterprises in the design
process, with methods currently available in the toolbox, and will be evaluated by these end-users
as well.
A toolbox supporting ICT design for specific user groups. End-user involvement low-
ers risks of new-product introduction and facilitates accessibility of ICT products and
services for all civilians.
A toolbox that helps you design for your specific user group and various design phases
by providing methods with step by step support. From preparation of the method to
analysis of the results.
A toolbox that enables redesign of current products and design for new end users.
Stimulating innovation that is really needed and used built on a solid scientific knowl-
edge base.
A toolbox that presents existing theoretical knowledge on inclusive design methods in
such a way that SMEs can apply it to designing their products, and provide practical
suggestions for improvement.
Human Computation
Jacomien de Jong
jacomien.dejong@tno.nl
Carien Caljouw
c.caljouw@eaglescience.nl
include.eaglescience.nl/
COMMIT/ project

A
lthough social applications such as Twitter, Foursquare,
Life360 and Facebook help to connect people, they can
conflict with important personal values like privacy, inde-
pendence and freedom. For example, when parents use
GPS-tracking to know where their children are 24/7, this
may be beneficial for family security, but it negatively impacts
children’s privacy and independence.
We have developed two mobile apps for increasing social par-
ticipation of people in the neighbourhood while at the same time
supporting users’ privacy, independence and freedom. The first
app is designed for children between nine and twelve years old,
to support them in their everyday activities (going to school,
making new friends, etc.). The second app is designed for el-
derly people and their social network to make them feel safer
and less dependent on others.
Can the use of norms in social applications promote user values like independence better than the
basic settings of existing applications like Facebook? How can social applications be designed to
make them more usable to the special needs of vulnerable people such as children and elderly?
Application
The application for children allows parents and children to ‘check-in’ at geographical locations they
define and share these check-ins with selected family and friend groups. The app provides the
ability to create customized agreements between users concerning the behaviour of the applica-
tion, based on a normative model. Parents can for example agree with their children under which
conditions check-ins are shared.
The application for elderly allows them to define the social network and to choose what information
(e.g. location) will be shared and via which medium (e.g. SMS). The app also provides the possibil-
ity to set up a scenario. For instance, in case of emergency, the elderly can send out an alarm to
49. Socially adaptive apps
the social network that automatically alarms the relevant persons, depending on the rules that the
elderly has defined.
The concepts of both our apps can be applied to other demanding target groups. For example,
elderly citizens can use the app designed for children to organize events and find caregivers clos-
est to their location. The concept of the social network alarm for elderly can be used by other target
groups as well, e.g. people living on their own.
Nice to know
The company LG is gearing up to launch a wristband that plots a child’s every move via a built-in
GPS sensor. The wristband is expected to launch this year in South Korea, and in Europe and the
United States one year later. The British newspaper The Guardian says that this sort of technology
will “edge us closer to a dystopian future”.
Smart social applications for wellbeing of vulnerable people such as children and
elderly.
Applications that can adapt to users’ values, promoting, e.g., safety and social connect-
edness, without harming others such as privacy and independence.
Making mobile social platforms smarter through novel adaptive technology to better
support people in their daily lives.
Socially adaptive context-aware applications to provide personalized support to people
in their daily lives.
Human Computation
Alex Kayal (App for children)
a.kayal@tudelft.nl
Jacomien de Jong (App for elderly)
jacomien.dejong@tno.nl
www.youtube.com/watch?v=xaAkU2MemQ4&feature=youtu.be
COMMIT/ project

A
t present there are about three hundred
million patients with diabetes worldwide.
Stress has a high impact on the suc-
cessful management of diabetes. Field
research has shown that diabetic people
actually get additional stress by the device they
use to check their blood sugar level. Stress re-
leases more sugar into the bloodstream, thus
creating a vicious circle. The current devices
focus strictly on quantification, leaving no room
for personal feeling.
We create a more intuitive interface to check the blood sugar level, so that the patient gains more
tranquillity. Instead of numbers, we use the colour spectrum of a LED-light to expresses the blood
sugar level. This more intuitive and sensitive language allows people to relate easier to their blood
sugar level while staying in touch with their own sensation. We call our concept ‘Measuring less to
feel more’.
How can we design pleasurable personal devices, which provide a clear feedback?
Application
The artificial pancreas will be clinically tested in the beginning of 2015, for CE certification. The
remaining question is: How to develop a glucose meter for taking blood-glucose measurements
that is a more personal device than present alternatives? Important design requirements are that
it should be pleasurable to use, should provide feedback relevant to the user and should diminish
stress.
The ‘Measuring Less to Feel More’ concept is further developed in close collaboration with Inreda
Diabetic BV. This is a Dutch company that develops an artificial pancreas. The artificial pancreas
50. Diabetes: Measuring Less
to Feel More
is a fully automated system that regulates a person’s blood sugar level via infusions. The main ad-
vantage is that the quality of life can be improved considerably. Diabetic patients have full diet and
exercise freedom. In addition, less finger pricks are necessary than with a regular treatment; only
for the calibration of the sensors.
Waag Society is supporting Inreda Diabetic BV with the design of the device with a special focus
on the interaction between the device and its user.
Nice to know
It is possible to take a shower with the artificial pancreas. It is also possible to temporarily discon-
nect the system when taking a shower.
It is possible to swim with the artificial pancreas. The device is water resistant and will not break
while swimming with it. However, the sensors cannot send their signals to the device while being
under water.
Quote
“If it works, I can feel almost Diabetes-free.”
“The artificial pancreas INREDA Diabetic regulates blood sugar automatically. Dia-
betic patients feel they are being ‘patient-off’.” (Herman Wijffels Innovation Public
Award 2013) The interaction design of the device is important.
For hundred millions of diabetes patients worldwide, the artificial pancreas could be an
outcome. It is a fully automated system that regulates a person’s blood sugar level via
infusions. The quality of life can be improved considerably.
The combination of an artificial pancreas with an intuitive and user-friendly interface,
which causes less stress in operating the device, will support diabetes patients to lead
a life as normal as possible.
“Scientists worldwide worked for years on an artificial pancreas, but still without much
success. We are the first to use glucagon besides insulin. Therefore our device resem-
bles much more secure a real pancreas.” (Robin Koops, founder Inreda Diabetics).
Human Computation
Esther van der Bijl
Esther@waag.org
Sabine Wildevuur
sabine@waag.org
COMMIT/ project
Inreda Diabetic BV

I
n 2013 TNO-research indicated that one
million people of the Dutch workforce
show signs of burnout and that stress is
the main reason for seven percent of all
sick leaves. With the goal of reducing
these numbers, we have developed the Fish-
ualization monitoring system. Fishualization
enables employees to gain insights into their
working habits to reduce stress and increase
productivity.
Fishualization is based on two components.
First, by combining multiple sensor data we au-
tomatically measure the activity levels, the amount of focus and the mental energy of workers.
Second, we show visualizations of the state or mood of an entire team of knowledge workers. Each
employee is represented as an avatar in the form of a fish. At the bottom of the screen we show
‘plants’ that represent a group of tasks, for example: writing e-mail, editing a document, web brows-
ing or preparing a presentation.
How to analyze and interpret heterogeneous multi-scale sensor data? What is a reliable model to
measure the state of an entire group of knowledge workers?
Application
The Fishualization feedback system can help to reduce stress and increase productivity at work on
the basis of an estimation of the workload and the mental and physical fitness of a worker.
The initial model relies mostly on computer interaction, identification of tasks, and context switches.
In the near future it will be combined with affect and physical aspects. We can also include the
analysis of facial expressions or e-mail sentiments.
51. Monitoring moods of workers
reduces sick leaves
At present there are no sensing based personal coaching products on the market that combine
multiple sensors and look at both mental and physical fitness.
Sensor-based reasoning and decision-making can also be deployed in smart cities or other smart
contexts. However, our Fishualization platform has the largest potential in other person-related
monitoring and coaching situations, such as self-management of people with chronic diseases or
e-learning systems.
Nice to know
“Studies conclude that contemplation of fish in an aquarium seem to have a significant effect in
reducing levels of stress and anxiety.”
“In Feng Shui, moving water is considered beneficial in balancing chi, and a well maintained aquar-
ium in the right location increases wealth and luck.”
The goal of the Fishualization golden demo is to enable office workers to gain insights
into their working habits in order to reduce stress levels, prevent burnout and increase
productivity.
Fishualization is a reconfigurable intervention aimed at improving well-being at work. It
is easily extended with new sensors, reasoning and mapping between data/information
and visualized fish behaviour.
Fishualization raises awareness regarding workload and related stress by visualizing
office worker activities as avatars on a centrally located display, which gives employees
of an entire department feedback.
Fishualization is a unique social intervention that provides a group visualization of the
estimated human computer activity and mental condition of a team of office co-workers
using a combination of various unobtrusive low-level sensors.
Human Computation
John Schavemaker
john.schavemaker@tno.nl
www.swell-project.net/
www.fishualization.nl
COMMIT/ project
SWELL Smart Reasoning Systems for Well-being at Work and at Home

W
e have developed the SWELL e-coaching app that helps knowl-
edge workers in achieving personal goals related to their work-life
balance. The app runs on a smartphone and can access a wide
variety of sensors that recognize physical activity, work activity and
working tasks.
Before using the e-coaching app the user is screened on physical fitness.
Via a questionnaire and a condition test personal parameters and goals
are evaluated. Subsequently, a user may enter a coaching program to
improve sleep, physical fitness or mental fitness.
Depending on the goal, the app will measure parameters, compare with
targets and provide feedback and suggestions. Key issue is that the app
avoids unsuitable moments for feedback and that it adjusts the feedback
according to the personality of the user.
How to reason on the basis of uncertain information inputs that come from very different types of sen-
sors? The sensors used are not proprietary developed sensors, but are off-the-shelf generic products.
Currently, many users stop using feedback apps because they are not enough personalized and
because the apps are too little aware of when to give feedback or not. We try to solve this problem
by building a coaching app that predicts the suitable moments of feedback for each user.
Application
Recent research (TNO, 2013) indicates that out of the Dutch workforce of 7.4 million people, one
million workers show symptoms of burnout. The same research identifies stress and workload as
the main reasons for at least seven percent of reported sick leaves. The total costs are estimated
to be nine hundred million euro per year.
At present there are no personal coaching products on the market that combine multiple sensors
and look at both mental and physical fitness.
52. Coaching-app helps to find
work-life balance
Jan Geert van Hall
jangeert@almende.org
COMMIT/ project
Our technology partners are Philips, Almende, Sense-OS. Competitors such as Fitbit have a single
sensor approach and are only focused on the lifestyle market.
Sensor-based reasoning and decision making is important to modern life. It can also be deployed in
smart cities or other smart contexts. The SWELL platform has the largest potential in other person
related monitoring and coaching situations, such as self management of people with chronic dis-
eases, or e-learning systems. Especially an adequate context aware intervention and messaging
algorithm would have a wide field of application for apps providing coaching and personal messag-
ing in any field.
Nice to know
Sense-OS will build a commercial pilot along the lines of the SWELL e-coaching app for ASML to
reduce and avoid burn-out with ASML employees.
Quote
“ We should do more with this!” − Reaction to SWELL presentation on context sensing; Well-being
@ work conference, June 2014.
The goal of the e-coach is to do a holistic health related analysis of the work and life-
style of office workers and to provide personalized activity recommendations in order
to prevent burn-out and reduce the cost of sick leave.
With the e-coach we will develop a generic coaching engine that, for the first time, will
adapt to personal needs and takes input from many (sensor) sources.
The proposed e-coaching app will decrease stress on the company workforce by hav-
ing a unique personalized and integrated approach whilst securing privacy of employ-
ees.
In the e-coach app, we try to combine knowledge of interaction between the physical
and mental aspects of a person, as to how these factors contribute to an over-all sense
of well-being or lack thereof.
Human Computation

G
athering objective data about every day life and work be-
haviour can help people to gain a better insight in both
harmful and helpful patterns in their lives. With this aim we
have developed the SWELL lifelog dashboard.
Using self-learning algorithms the SWELL lifelong dash-
board combines data from various unobtrusive sensors, includ-
ing workplace, smartphone and body worn sensors. All data
are stored in a personal cloud store and can be downloaded or
shared with others. Based on these measurements, the dash-
board gives users insights into various physical, mental and work variables over a long period of
time.
How can we unobtrusively track physical and mental well-being, both at work and at home? Which
algorithms are best suited for this task? In which way do we have to display the information so that
people make optimal use of the dashboard?
Our approach is unique in its focus on individual users, its flexible set of different sensors, its novel
algorithms.
Application
Recent research (TNO, 2013) indicates that fourteen percent of the entire Dutch workforce shows
symptoms of burnout and that stress related sick leaves account for an estimated cost of roughly
nine hundred million euro per year. In an unique holistic combination of mental and physical states,
by quantifying individual work and life, our SWELL lifelong dashboard can contribute to a better
well-being.
Our technology partner Sense OS is currently commercializing different parts of the technology for
several launching customers within her Sense Health daughter company.
53. Quantifying your life for
a better well-being
Sensor based reasoning and decision-making can also be applied in smart cities or smart contexts.
However, the SWELL platform has the largest potential in other person related monitoring and
coaching situations, such as self-management of chronic diseases, e-learning systems, serious
games, mental healthcare, or personal security. Because of the modular architecture of our tech-
nology, applying parts of it in other contexts will be relatively easy.
Nice to know
ASML Corporate Vitality Manager Maaike Thijssen has won the Health Manager 2014 award based
on her employee vitality program with the technology developed in the SWELL context. ASML is a
launching customer for Sense’s vitality solution.
Our technology partner Sense has won an EU eHealth award 2013 for a mental health application
incorporating some of the SWELL technology.
Quote
At the international well-being at work conference, SWELL was the only project in which data was
sensed for personalized coaching. Many reactions were like: “We should do more with this!”
Improving creativity and productivity of knowledge workers through effortless tracking
and feedback of mental and physical well-being.
Real-time self-learning algorithms combine data from 15 different unobtrusive sensors,
(workplace, smartphone, wearables) to infer user-centric state information like sleep,
activity, valence, stress, and work task.
Low cost user-centric solution for effortless monitoring of stress and burnout symptoms
to help users prevent absenteeism and improve productivity of knowledge workers.
Validated algorithms for unobtrusive tracking and visualizing of both physical and men-
tal well-being, and work context, without the need for proprietary hardware.
Human Computation
Joris Janssen
joris@sense-os.nl
swelldemo.commonsense-dashboard.com
www.commonsense-dashboard.com
COMMIT/ project

M
odern medicine tries to understand diseases
more and more by looking at the molecular fin-
gerprint of a disease. This is done by molecular
imagingofbiologicaltissues.Molecularimaging
can assist in the diagnosis and prognosis of dis-
eases. It also enables the development of medicine
specific for you and not just for everybody, a boom-
ing research field called ‘personalized medicine’.
Our demo presents a 3D-visualization of molecular
imaging data generated by Mass Spectrometry Im-
aging (MSI). MSI is a technique for the simultaneous
detection and visualization of a large variety of mol-
ecules based on their molecular masses. Using our
interactive tool you can view and explore 3D-images
of molecular breast cancer tissue.
The main scientific challenge is to reduce, process,
analyze and interpret huge datasets.
Application
Molecular imaging with mass spectrometry easily generates large amounts of complex data. The
embedding of molecular imaging in a clinical setting requires easy data visualization and extraction
of relevant information.
In our demo you will experience the complexity of molecular imaging data. We expect that the
information extracted from this data will assist medical doctors in clinical decision making in the
future. It can then be used for example for diagnosis, prognosis and treatment response prediction.
54. Molecular biobanks unravel
the secrets of breast cancer
Alternative applications may lie in the field of different medical imaging modalities like MRI or
microscopy. MRI or microscopy data can be correlated with molecular imaging from mass spec-
trometry.
A completely different possible application lies in the field of the molecular analysis of different lay-
ers of paint that compose for example a Van Gogh- or a Rembrandt-painting.
Nice to know
One molecular image can contain more than one million spectra.
Visualization of molecular data for personalized, predictive, participatory and preven-
tive medicine.
The development of software tools that can analyze complex molecular imaging data
brings us closer to personalized medicine, the future of patient health care.
Innovative medical image generation, data processing and visualization (in cancer re-
search).
Interactive 3D visualization and analysis of complex molecular imaging data.
Human Computation
Nadine Mascini
N.Mascini@amolf.nl
www.amolf.nl
COMMIT/ project
ps-tech

U
until now radio content has been disconnected from other audiovisual and textual informa-
tion. Our demo shows that radio can be much more then listening only. Radio can be made
searchable to find a certain quote, person or topic. The experience of listening radio can be
enhanced presenting related images from other information sources on a screen.
Reversely, radio content can also be linked to information on the web. A written news story
might be enriched with a related radio interview.
We improve the use and experience of radio content by automatically processing audio files and
combining the output with the latest audio search technology.
How can information technology add value to media by enabling concepts that increase the use of
audiovisual content?
The ICT-solution should look for a combination of speech technology and information extraction
to decode the spoken words automatically and use these as labels for searching and linking func-
tionalities.
Application
How to enhance the experience of users engaging with the content? How do users evaluate these
new functionalities and how can this feedback be used to fine-tune the technology?
We have developed a lab version of a platform for online radio, enhanced with automated content-
based search, browsing and linking. We are also pleased to connect our technology to a potentially
large user base via the Woord.nl platform: a collection of historical and recent radio stories.
Partners in our work are media industry representatives: content-owners (VPRO), archives (NIBG)
and technology providers (Cross-Media Interaction).
Similar applications have the disadvantage that they require substantial manual resources to
accomplish the same functionalities Therefore they are limited in scope.
55. Enriching use and experience
of radio content
Roeland Ordelman
rordelman@beeldengeluid.nl
woord.nl
COMMIT/ project
Human Computation 116 117Human Computation
The ICT science that we develop is applicable to any domain that uses audiovisual content.
Alternative applications might therefore lie in the domain of meetings, presentations, talks, gaming
etc.
Nice to know
Our technology is implemented in a lab version of the woord.nl portal; it is used at the Nether-
lands Institute for Sound and Vision for searching audiovisual collections; and it is also adopted
in a search application for researchers in the Digital Humanities interested in Oral History content
(http://zoeken.verteldverleden.nl)
RadioPlus demonstrates audiovisual search to aid in the browsing of audio collections
from diffuse sources. For industry, the technology enables education, journalism, re-
search, and security.
For the holder of digital archives, better means are needed to exploit the content of
the archives. RadioPlus connects audiovisual content with other content to create new
engaging access tools for users.
RadioPlus provides a scenario for the impact of audiovisual search and linking technol-
ogy for the use of multimedia content. The role of audiovisual media in every day life
pleads for such technology.
The work behind RadioPlus aims to enhance search and exploration of large spoken
word collections and to enable cross-media audio hyperlinking by spoken entities de-
rived from the audio signal.

C
ultural heritage institutions more and
more provide their collections online. To
make them accessible to a interested
users, the collections need short de-
scriptions (annotations). Creating such
annotations requires knowledge and expertise
that is not always possessed by the collection
curators. The usage of crowdsourcing and nich-
esourcing techniques provides cultural herit-
age institutions with new tools to create high-quality annotations in an inexpensive and fast way.
We demonstrate Accurator, a platform for the enhancement of the annotation process of cultural
heritage institutions like museums and libraries. The enhancement is done with the domain-specific
expertise of amateur enthusiasts and experts drawn from the crowd.
We show an demo of the Accurator platform that features innovative, easy-to-use content annota-
tion applications, driven by domain-specific knowledge. These applications include a novel Web
annotation interface; an expert finding module for social media and Q&A-forums; and an online
dashboard for annotation campaigns monitoring and control.
How to design efficient and effective crowdsourced content annotation is still an open research
question. With our work we seek answers to the following research questions: How can large
crowds and expert niches be activated to share their knowledge? How can expertise be identified
on the Web? How can experts be discovered and engaged? How can linked data support and drive
the content annotation process? How can semantically enriched knowledge repositories improve
search and recommendation applications?
Application
Our work is driven by the needs of public organizations such as the Rijksmuseum, the Naturalis
Biodiversity Center, the National Library, and the Museum Het Prinsenhof in Delft. The Accurator
56. Nichesourced annotation of
cultural heritage collections
Alessandro Bozzon
a.bozzon@tudelft.nl
sealincmedia.wordpress.com/
COMMIT/ project
platform provides methods and tools to improve the quality of digital content annotation, and, ulti-
mately, the performance of content exploration and retrieval tools.
We demonstrate the benefits ofAccurator on the search interface of the Rijksmuseum Website, where
high-quality crowdsourced annotation directly leads to improved search and exploration experience.
In comparison with existing platforms for content annotation Accurator advances the state of the
art by providing content curators with knowledge-driven tools for expert retrieval, evaluation, and
engagement.
The Accurator platform can be easily transformed for usage in other domains. This is demonstrated
by many on-going collaborations with companies such as IBM (knowledge creation and retrieval,
software engineering) and Frontwise (event-centric search for audio-visual archive).
Nice to know
Experiments have shown that the collective knowledge of the crowd matches at least one-third of
the knowledge of domain experts.
Large amount of crowd workers can be easily engaged in cultural heritage content annotation
tasks. For example, around one thousand workers were recruited in just a couple of days for the
annotation of Rijkmuseum prints.
Support content curators in museums and similar professional organizations in their
campaigns using annotations created by pro-amateurs.
Push the boundaries of crowdsourcing to efficiently valorize the passion, knowledge,
and skills of crowds.
Enable customized expert annotations for next-generation multimedia content retrieval
and fruition platforms.
Exploit crowd sourced content annotations for professional high-quality annotation
tasks.
Human Computation

R
esearchers, publishers, and funding agen-
cies increasingly recognize the impor-
tance of publishing the original research
data along with traditional journal articles.
However, the threshold for publishing data
in a way that enables reproducibility and reuse,
is still too high for an unsophisticated scientist.
To solve this problem, we have developed a
web-based dashboard which incorporates a
number of techniques for enrichment of re-
search data with appropriate metadata, such as
links to relevant external resources and identi-
fiers. It also helps the user to upload the results to a popular research data repository were the data
can be discovered, verified, and ultimately reused by other researchers.
How can scientists publish their research data in such a way that their colleagues can easily repro-
duce or reuse them?
We approach this problem by automating metadata discovery and enabling the user to publish
results as Linked Data following standards and best practices. For the convenience of the user we
hide the details of the underlying semantic web technologies.
Application
We apply the technology to support researchers who are not necessarily data management ex-
perts, but produce and publish original scientific data. A number of related efforts aiming at low-
ering the barriers for data publication and improving the quality of published data are underway.
However, these efforts are platform-dependent and do not comply with emerging standards for
57. One-click semantic enrichment
of scientific data
Marat Charlaganov
m.charlaganov@vu.nl
www.data2semantics.org
COMMIT/ project
data publishing. Our approach relies on the latest semantic web standards for data publication and
enables easy automatic processing of published results.
The exploration of texts is broadly needed. Rather than rely on researchers for enriching metadata
and publishing Linked Data, repositories might use our technology in their user-facing applications,
and for batch processing of existing deposits. We are collaborating with Mendeley (Elsevier’s free
reference manager and academic social network) and with Figshare (an online digital repository
where researchers can preserve and share their research outputs) in order to bring parts of our
technology to their products.
Nice to know
Metadata is structured information that describes, explains, locates, or otherwise makes it easier
to retrieve, use, or manage an information resource. Metadata is often called data about data or
information about information.
The value of scientific data is determined by reuse in academia and industry. Therefore
every effort should be made to make all research data available and discoverable.
There is only one-way to ensure interoperability with future applications and services:
follow the standards!
In the world where high quality scientific data is easy to find and verify innovation ac-
celerates tremendously.
Data publication must provide enough context to be interpreted and allow citation. It’s
not that different from a journal publication, after all.
Human Computation

I
CT-based home healthcare services are
revolutionizing healthcare. The patient
performs his own medical measurements
and sends the results via the internet to the
healthcare professional. The doctor has to trust and rely on the measurements to provide the
correct medical advice. Unfortunately, people sometimes use a measurement device incorrectly, so
that the reliability of the data becomes questionable.
Our demo assesses the overall quality of medical measurements by evaluating various quality
aspects. We have developed a novel troubleshooting tool that can find possible causes of low data
quality. For example, a patient’s unstable measurement can be the result of incorrect sensor ap-
plication, but it can also be the result of health deterioration.
Our technology provides the doctor with a unique interpretation of the patient’s own medical meas-
urements. For the acceptance of home healthcare services data reliability is crucial
How to evaluate the overall quality of measurements? The evaluation should involve different qual-
ity aspects of a measurement, such as information stability, device application and data timeliness.
Application
Our approach evaluates the measurement quality of medical measurements that people take at
home by an activity monitoring system. Patients sometimes misuse the activity sensor and provide
unstable measurements which complicates the interpretation by the doctor. In particular, patients
might forget to apply the device or they place it incorrectly. Our application is able to detect such
miscues.
With our partner, the Roessingh rehabilitation centre, we evaluate the activity measurement of
COPD patients.
58. Evaluating the reliability
of eHealth-data
Sokratis Vavilis
S.Vavilis@tue.nl
security1.win.tue.nl/~svavilis/
COMMIT/ project
Our approach can be applied to other domains that heavily rely on sensor data. Interesting alterna-
tive domains for application could be the automotive industry and smart energy grids, where multi-
ple sensors are used and a unique interpretation is needed. Moreover, our approach can be used
to investigate the causes of low data quality.
Reliable data contribute to fewer hospital visits, better diagnoses, lower costs; overall,
a better healthcare.
Methods for evaluating data reliability can be integrated into healthcare services as
plug and play components.
Methods enable doctors to trust and to rely on remotely measurements. Novel trouble-
shooting mechanisms are used to obtain the causes of low quality measurements. In
this way patient compliance is improved.
The system includes methods to assess and to integrate various quality aspects of a
measurement, such as information stability, measurement procedure and data timeli-
ness.
Human Computation

A
growing number of users demand from
service providers to offer transparency on
how their data are handled. Data trans-
parency is also more and more demanded
by the law.
We have developed a toolkit with various pri-
vacy transparency tools that can be used in
e-Health services, such as a hospital website.
From our toolkit, service providers can choose
specific tools that fit their user requirements and
their service type the best. Our tools provide users with insight in how service providers claim to
handle their sensitive data, how their sensitive data are actually handled, and with which third par-
ties the data are shared.
The use of privacy transparency tools will lead to higher trust by users and hence a higher service
acceptance. Ultimately this results in more available data and better medical care.
How can we best design tools that give users of information services insight in how their data are
being handled? The tools should provide the right type and amount of information and in the right
form, so not to overwhelm an average user.
The technical challenge is to provide complete information on what service providers actually do
with data and why, not only what they claim to do (such as e.g. Google dashboard).
Application
In the area of e-Health, our demonstration shows an implementation of privacy transparency tools
for the case of the COPD portal of the Roessingh rehabilitation centre.
What makes our solution different from similar ones is that in the existing tools transparency is a
side effect, and not a tool’s main purpose. Existing tools have limited functionalities and are not
59. Privacy transparency tools
give more trust in eHealth
customized for different types of users and services. Also none of the tools has been validated on
the effect.
Our solution is not only easily applicable to different types of eHealth services, but also to other
service domains, for example e-Government, e-Commerce, e-Finance etc.
Nice to know
Dutch law dictates parties who process personal data to actively inform their customers on the
data processing they do. It is illegal to hide what you do with personal data or to only tell it when
customers ask for it.
Quote
“Transparency about personal data fits in with our principle of transparency about providing care
in general.”
Enhanced transparency increases end-user trust for organisations offering online ser-
vices.
Transparency tackles the increasing societal concerns about personal data usage.
An organization’s vision on honest handling of personal data is not a hollow promise:
transparency tools make it concrete.
The transparency tools are an answer the question of which type of transparency in-
formation to present to the end-user, at which level of detail, and how to present it to
maximize end-user.
Human Computation
Milena Kooij-Janic
milena.kooij@tno.nl
Jan Pieter Wijbenga
jan_pieter.wijbenga@tno.nl
COMMIT/ project

COMMIT/ projects and project leaders 126 127COMMIT/ projects and project leaders
www.commit-nl.nl/projects/information-retrieval-for-information-services
Prof.dr. Maarten de Rijke
M.deRijke@uva.nl
www.commit-nl.nl/projects/interaction-for-universal-access
Prof.dr.ir. Dirk Heylen
d.k.j.heylen@utwente.nl
www.commit-nl.nl/projects/sensei-sensor-based-engagement-for-improved-
health
Mr. Vikas Kannav
Vikas_kannav@infosys.com
www.commit-nl.nl/projects/virtual-worlds-for-well-being
Drs. Albert Vlug
albert.vlug@cgi.com
www.commit-nl.nl/projects/socially-enriched-access-to-linked-cultural-media
Prof.dr. Alan Hanjalic
A.Hanjalic@tudelft.nl
www.commit-nl.nl/projects/swell-smart-reasoning-systems-for-well-being-at-
work-and-at-home
Prof.dr.ir. Wessel Kraaij
wessel.kraaij@tno.nl
www.commit-nl.nl/projects/sensor-networks-for-public-safety
Prof.dr.ing Paul Havinga
P.J.M.Havinga@utwente.nl
www.commit-nl.nl/projects/very-large-wireless-sensor-networks-for-well-
being
Prof.dr. Maarten van Steen
steen@cs.vu.nl
www.commit-nl.nl/projects/composable-embedded-systems-for-health-
care
Prof.dr. Jozef Hooman
jozef.hooman@esi.nl
www.commit-nl.nl/projects/metis-dependable-cooperative-systems-for-
public-safety
Mr. Dave Watts
dave.watts@tno.nl

Colofon
The Board of COMMIT/
Arnold Smeulders
Peter Apers
Inald Lagendijk
Geleyn Meijer
Johan Vos
Produced by the office of
COMMIT/
Mieke van den Berg
Tjerk de Greef
Editor: Bennie Mols
Photos are provided by the
respective demo owners
Photo introduction:
Jan Willem Steenmeijer
Graphic design:
Studio Ron van Roon
Printer: Drukkerij Terts
© COMMIT/ October 2014
4COMMIT/ projects and project leaders
www.commit-nl.nl/projects/trusted-healthcare-services
Prof.dr. Milan Petkovic
milan.petkovic@philips.com
TimeTrails Spatiotemporal Data Warehouses
for Trajectory Exploitation
www.commit-nl.nl/projects/spatiotemporal-data-warehouses-for-
trajectory-exploitation
Prof.dr. Martin Kersten
Martin.Kersten@cwi.nl
www.commit-nl.nl/projects/e-infrastructure-virtualization-for-e-science-
applications
Prof.dr.ir. Henri Bal
bal@cs.vu.nl

Data2Semantics From Data to Semantics
for Scientific Data Publishers
www.commit-nl.nl/projects/from-data-to-semantics-for-scientific-data-
publishers
Prof.dr. Frank van Harmelen
Frank.van.Harmelen@cs.vu.nl
e-FOOD e-Foodlab
www.commit-nl.nl/projects/e-foodlab
Prof.dr. Jan Top
jan.top@wur.nl

0872387293638
1505782012395
8692217824301
6278249526532
9495782014630
4924629726045

Def COMMIT Demoboekje V2 LR

More Related Content

What's hot (16)

Viewers also liked (6)

Similar to Def COMMIT Demoboekje V2 LR (20)

Def COMMIT Demoboekje V2 LR