SlideShare a Scribd company logo

Ph. D. Final Dissertation SLides

Emanuele Panigati, profile picture
Emanuele Panigati

This document describes Emanuele Panigati's doctoral dissertation on the SuNDroPS system for managing semantic and dynamic data in pervasive systems. It provides an overview of SuNDroPS and its components for processing streaming and historical data, including Context-ADDICT for querying heterogeneous data sources and PerLa and Tesla for information flow processing. It also describes how SuNDroPS was tested in the motivating Green Move vehicle sharing scenario.

Presentations & Public Speaking
1 of 43
Download to read offline
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
30
31
32
33
34
35
36
37
38
39
40
41
42
43
SuNDroPS: Semantic and dyNamic Data in a Pervasive System Context-ADDICT Revisited Doctoral Dissertation of: Emanuele Panigati Advisor: Prof. Letizia Tanca Co-Advisors: Prof. Fabio A. Schreiber, Prof. G. Cugola Politecnico di Milano { Dipartimento di Elettronica, Informazione e Bioingegneria December 10th, 2014
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions & Future Works Summary of the content Introduction and Motivation Go A motivating scenario: the Green Move project Go The SuNDroPS system Go The SuNDroPS legacy blocks Go Context-ADDICT Go PerLa & Tesla Go Hystorical data analysis Go The SuNDroPS new components Go New features of PerLa Go New features of Tesla/TRex: SemTRex Go MR-Miner & MREClaT Go Testing SuNDroPS in the Green Move scenario Go Conclusions and
nal remarks Go E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions & Future Works Introduction and Motivation Users are surrounded by a high quantity of heterogeneous data, often in the form of data streams Humans cannot fully exploit the whole richness of these data without digital support for their analysis Real-time, on-the- y and historical data processing are equally necessary to obtain useful knowledge E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions & Future Works A Motivating Scenario: the Green Move Project Green Move is a zero-emission vehicle-sharing system which supports the users with additional digital services The Green Move project has been used as a real-world on-
eld test, using several SuNDroPS components The user experience is
nely personalized based on the user context and on contextual preferences, so the data management process must consider a contextual tailoring of the data Data coming as streams from dierent kinds of sensors (e.g., on-board vehicle status sensor, environmental sensor, . . . ) must be processed on-the- y in order to give to users an immediate feedback and empower their experience. Dierent Information Flow Processing systems have been considered to perform this task (PerLa TRex) E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis SuNDroPS Architecture Overview SuNDroPS allows to manage the ow of (possibly semantically enriched) information contained in data streams and to study how to extract useful knowledge from it and from the more traditional, static datasets. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis SuNDroPS Architecture Overview SuNDroPS allows to manage the ow of (possibly semantically enriched) information contained in data streams and to study how to extract useful knowledge from it and from the more traditional, static datasets. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Legacy: The Context-ADDICT system Allows to query dierent and heterogeneous data sources providing a single entry-point for queries Automatically tailors the query according to the current context of the user, and rewrites it, integrating the results coming from each dierent data source E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Legacy: PerLa Tesla Two information ow processing systems PerLa is based on a DSMS paradigm Tesla/TRex is based on a Complex Event Processing paradigm E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis More on Information Flow Processing Two dierent approaches: DSMSs, developed by the database community, consider a data stream as a sequence of tuples, processing them using SQL-like query languages CEPs, developed by the distributed software engineering community, consider the stream as a sequence of events and process them using rule and/or logic based languages for temporal pattern detection PerLa is an example of the
rst kind of systems while Tesla/TRex belongs to the second category. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis More on Information Flow Processing Two dierent approaches: DSMSs, developed by the database community, consider a data stream as a sequence of tuples, processing them using SQL-like query languages CEPs, developed by the distributed software engineering community, consider the stream as a sequence of events and process them using rule and/or logic based languages for temporal pattern detection PerLa is an example of the
rst kind of systems while Tesla/TRex belongs to the second category. We next compare the features of the two approaches. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { PerLa Data VehicleData greenBox id Timestamp Speed A300A 10/12/2014 14:00 0.0 B400B 10/12/2014 14:15 15.0 C100C 10/12/2014 14:01 50.0 TakenOrReleased greenBox id Timestamp takenReleased A300A 10/12/2014 7:00 TAKEN A300A 10/12/2014 8:00 RELEASED B400B 10/12/2014 10:00 TAKEN B400B 10/12/2014 11:00 RELEASED C100C 10/12/2014 12:00 TAKEN E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { PerLa Queries PerLa Low Level Query CREATE SNAPSHOT MostRecentUse (greenBox id String, takenReleased Integer, date[3] Integer) WITH DURATION 10 AS LOW: EVERY 30 m SELECT greenBox id, takenReleased, date[3] HAVINGdate = MAX(date, 10) UP TO 30m SAMPLING ON EVENT takenInCharge Released PerLa High Level Query CREATE OUTPUT STREAM Theft (greenBox id String, recentUsage date) AS HIGH: EVERY 10 m SELECT greenBox id, MAX(MostRecentUse.date) as mass FROM MostRecentUse, TakenOrReleased, VehicleData WHERE VehicleData.greenBox id = TakenOrReleased.greenBox id AND TakenOrReleased.date = mass AND TakenOrReleased.takenReleased = 0 AND VehicleData.speed 0 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { PerLa Query Results MostRecentUse greenBox id takenReleased date A300A 10/12/2014 8:00 RELEASED B400B 10/12/2014 11:00 RELEASED C100C 10/12/2014 12:00 TAKEN Theft greenBox id RecentUsage B400B 10/12/2014 11:00 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { Tesla Data Rule Events event id : 17; greenBox id : A300A; ts : 10=12=2014 14 : 00; speed : 0:0 event id : 17; greenBox id : B400B; ts : 10=12=2014 14 : 15; speed : 15:0 event id : 17; greenBox id : C100C; ts : 10=12=2014 14 : 01; speed : 50:0 event id : 112; greenBox id : A300A; ts : 10=12=2014 7 : 00 event id : 121; greenBox id : A300A; ts : 10=12=2014 8 : 00 event id : 112; greenBox id : B400B; ts : 10=12=2014 10 : 00 event id : 121; greenBox id : B400B; ts : 10=12=2014 11 : 00 event id : 112; greenBox id : C100C; ts : 10=12=2014 12 : 00 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { Tesla Data Rule Events event id : 17; greenBox id : A300A; ts : 10=12=2014 14 : 00; speed : 0:0 event id : 17; greenBox id : B400B; ts : 10=12=2014 14 : 15; speed : 15:0 event id : 17; greenBox id : C100C; ts : 10=12=2014 14 : 01; speed : 50:0 event id : 112; greenBox id : A300A; ts : 10=12=2014 7 : 00 event id : 121; greenBox id : A300A; ts : 10=12=2014 8 : 00 event id : 112; greenBox id : B400B; ts : 10=12=2014 10 : 00 event id : 121; greenBox id : B400B; ts : 10=12=2014 11 : 00 event id : 112; greenBox id : C100C; ts : 10=12=2014 12 : 00 Rule DEFINE Theft (ID : String) FROM VehicleData (greenBox id = $id AND speed 0) AND LAST Release (greenBox id = $id) WITHIN 10day FROM VehicleData AND NOT Taken (greenBox id = $id) BETWEEN Release AND VehicleData WHERE ID = VehicleData.greenBox id E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { Tesla Data Rule Events event id : 17; greenBox id : A300A; ts : 10=12=2014 14 : 00; speed : 0:0 event id : 17; greenBox id : B400B; ts : 10=12=2014 14 : 15; speed : 15:0 event id : 17; greenBox id : C100C; ts : 10=12=2014 14 : 01; speed : 50:0 event id : 112; greenBox id : A300A; ts : 10=12=2014 7 : 00 event id : 121; greenBox id : A300A; ts : 10=12=2014 8 : 00 event id : 112; greenBox id : B400B; ts : 10=12=2014 10 : 00 event id : 121; greenBox id : B400B; ts : 10=12=2014 11 : 00 event id : 112; greenBox id : C100C; ts : 10=12=2014 12 : 00 Rule DEFINE Theft (ID : String) FROM VehicleData (greenBox id = $id AND speed 0) AND LAST Release (greenBox id = $id) WITHIN 10day FROM VehicleData AND NOT Taken (greenBox id = $id) BETWEEN Release AND VehicleData WHERE ID = VehicleData.greenBox id Results event id : 301; greenBox id : B400B; ts : 10=12=2014 14 : 15 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { PerLa Data VehicleData greenBox id GPS Speed A300A 45.1,15.1 30.0 B400B 45.1,20.2 15.0 C100C 42.2,15.1 50.0 Weather GPS Climate Limit 45.1,15.1 Normal 130 45.1,20.2 Rain 90 42.2,15.1 Ice 30 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { PerLa Queries PerLa Low Level Query CREATE STREAM WeatherChange (position gps data, climate String) AS LOW: EVERY 10 m SELECT position, climate SAMPLING ON EVENT WeatherChanged WHERE climate = Rain OR climate = Ice OR climate = Snow OR climate = Fog REFRESH EVERY 5 m PerLa High Level Query CREATE OUTPUT SNAPSHOT DangerousDriving (greenBox id String) WITH DURATION 2 h AS HIGH: SELECT greenBox id FROM VehicleData, WeatherChange, Weather WHERE VehicleData.gps data = WeatherChange.position AND WeatherChange.climate = Weather.climate AND VehicleData.speed Weather.limits E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { PerLa Query Results DangerousDriving position climate 45.1,20.2 Rain 42.2,15.1 Ice WeatherChange greenBox id C100C E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { Tesla Data Events event id : 17; greenBox id : A300A; pos : 45:1; 15:1; speed : 30:0 event id : 17; greenBox id : B400B; pos : 45:1; 20:2; speed : 15:0 event id : 17; greenBox id : C100C; pos : 42:2; 15:1; speed : 50:0 event id : 40; pos : 45:1; 15:1; climate : Normal; temp : 20 event id : 40; pos : 45:1; 20:2; climate : Rain; temp : 17 event id : 40; pos : 42:2; 15:1; climate : Ice; temp : 2 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { Tesla Rules Results Rule (Rain) DEFINE DangerousDrivingRain(ID : String) FROM VehicleData(speed90) AND LAST Weather(VehicleData.pos-xposVehicleData.pos+x) WITHIN 1h FROM VehicleData AND Weather.climate=rain WHERE DangerousDrivingRain.ID = VehicleData.greenBox id Rule (Ice) DEFINE DangerousDrivingIce(ID : String) FROM VehicleData(speed50) AND LAST Weather(VehicleData.pos-xposVehicleData.pos+x and temp0) WITHIN 1h FROM VehicleData WHERE DangerousDrivingRain.ID = VehicleData.greenBox id Results event id : 340; greenBox id : C100C E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Legacy: Hystorical Data Analysis Data Mining allows to extract knowledge from the gathered data, discovering previously unknown facts from them. Frequent Itemset Mining
nds in the database all the sets of items whose frequency is above a given support threshold Several algorithms are available to perform this task: A Priori Partition FP-Growth EClaT E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Legacy: Hystorical Data Analysis Data Mining allows to extract knowledge from the gathered data, discovering previously unknown facts from them. Frequent Itemset Mining
nds in the database all the sets of items whose frequency is above a given support threshold Several algorithms are available to perform this task: A Priori Partition FP-Growth EClaT E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT New Components in the Big Data Era SuNDroPS Adds new features to Context-ADDICT: Monitors the environment directly, using sensors, also reasoning on the gathered data Automatically infers (part of) the user context from the environmental data that have been sensed Integrates historical data processing with analysis operations, introducing a new parallel Data Mining algorithms E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT New features of PerLa The PerLa middleware has been completely reengineered to include asynchronous behaviors of sources (sensors, web services, . . . ) Distributed PerLa allows to exploit the sources (and network) computation power PerLa for Context explicitly integrates the context-aware approach of Context-ADDICT with Context-Oriented Programming COP, allowing sensors to behave dierently according to their current context E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT New features of Tesla/TRex: SemTRex Original TRex cannot interact with static data SemTRex adds a RDF static data repository to TRex and new operators in the Tesla language (IN) Integrating RDF repositories allows reasoning on the data IN allows to: Enrich the events, including into them facts retrieved from the KB Filter the events using facts included in the KB Prefetching and Caching of data become necessary to keep reasonable response times: basic cache, parametric basic cache, frequent data cache, combined caching strategy. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MapReduce-based Frequent Itemset Mining MR-Miner supports the mining processes in SuNDroPS using MREClaT, an EClaT-based algorithm exploiting the MapReduce programming paradigm, that allows SuNDroPS to analyze the data load typical of Big Data scenarios E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MapReduce-based Frequent Itemset Mining MR-Miner supports the mining processes in SuNDroPS using MREClaT, an EClaT-based algorithm exploiting the MapReduce programming paradigm, that allows SuNDroPS to analyze the data load typical of Big Data scenarios E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MR-Miner MREClaT { Algorithm Details First step: Mine 1-frequent itemsets E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MR-Miner MREClaT { Algorithm Details Second step: Mine 2-frequent itemsets E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MR-Miner MREClaT { Algorithm Details Third step: Mine k-frequent itemsets E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT Experiments E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT Pre
x Extension Experiments E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Testing SuNDroPS in the Green Move Scenario PerLa context-aware sensors SemTRex Pervasive and context-aware information push Context-aware vehicle assignment to user reservation, based on contextual user preferences E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works Conclusions The SuNDroPS system helps users to deal with the high information load that surronds them Context inference based on the environmental sensor data ows Historical data mining using parallel MapReduce algorithms to speed up processing Semantic-enhanced complex event processing using cache to reduce the performance degradation due to the disk bottle-neck Reengineering of the PerLa middleware allowing its distribution on the network components and the integration with other context-oriented paradigms (e.g. Context Oriented Programming) E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System

More Related Content

PPTX
Data Streaming in Big Data Analysis
Vincenzo Gulisano
 
PDF
Cloud-based Data Stream Processing
Zbigniew Jerzak
 
PDF
Scientific Application Development and Early results on Summit
Ganesan Narayanasamy
 
PPTX
The Pacific Research Platform: A Regional-Scale Big Data Analytics Cyberinfra...
Larry Smarr
 
PPTX
Taming Big Data!
Ian Foster
 
PDF
Dynamic Data Center concept
Miha Ahronovitz
 
PDF
The Interplay of Workflow Execution and Resource Provisioning
Rafael Ferreira da Silva
 
PPTX
Accelerating Discovery via Science Services
Ian Foster
 
Data Streaming in Big Data Analysis
Vincenzo Gulisano
 
Cloud-based Data Stream Processing
Zbigniew Jerzak
 
Scientific Application Development and Early results on Summit
Ganesan Narayanasamy
 
The Pacific Research Platform: A Regional-Scale Big Data Analytics Cyberinfra...
Larry Smarr
 
Taming Big Data!
Ian Foster
 
Dynamic Data Center concept
Miha Ahronovitz
 
The Interplay of Workflow Execution and Resource Provisioning
Rafael Ferreira da Silva
 
Accelerating Discovery via Science Services
Ian Foster
 

What's hot (15)

PPT
Grid optical network service architecture for data intensive applications
Tal Lavian Ph.D.
 
PPTX
Grid'5000: Running a Large Instrument for Parallel and Distributed Computing ...
Frederic Desprez
 
PDF
Scalable Distributed Real-Time Clustering for Big Data Streams
Antonio Severien
 
PDF
A Recommender Story: Improving Backend Data Quality While Reducing Costs
Databricks
 
PDF
04 open source_tools
Marco Quartulli
 
PPTX
Learning Systems for Science
Ian Foster
 
PPTX
Materials Data Facility: Streamlined and automated data sharing, discovery, ...
Ian Foster
 
PDF
07 data structures_and_representations
Marco Quartulli
 
PPTX
Discovery Engines for Big Data: Accelerating Discovery in Basic Energy Sciences
Ian Foster
 
PPTX
Accelerating Data-driven Discovery in Energy Science
Ian Foster
 
PPTX
Big data at experimental facilities
Ian Foster
 
PPTX
Challenges and Issues of Next Cloud Computing Platforms
Frederic Desprez
 
PPTX
Next Generation Grid: Integrating Parallel and Distributed Computing Runtimes...
Geoffrey Fox
 
PPTX
Opportunities for X-Ray science in future computing architectures
Ian Foster
 
PDF
Share and analyze geonomic data at scale by Andy Petrella and Xavier Tordoir
Spark Summit
 
Grid optical network service architecture for data intensive applications
Tal Lavian Ph.D.
 
Grid'5000: Running a Large Instrument for Parallel and Distributed Computing ...
Frederic Desprez
 
Scalable Distributed Real-Time Clustering for Big Data Streams
Antonio Severien
 
A Recommender Story: Improving Backend Data Quality While Reducing Costs
Databricks
 
04 open source_tools
Marco Quartulli
 
Learning Systems for Science
Ian Foster
 
Materials Data Facility: Streamlined and automated data sharing, discovery, ...
Ian Foster
 
07 data structures_and_representations
Marco Quartulli
 
Discovery Engines for Big Data: Accelerating Discovery in Basic Energy Sciences
Ian Foster
 
Accelerating Data-driven Discovery in Energy Science
Ian Foster
 
Big data at experimental facilities
Ian Foster
 
Challenges and Issues of Next Cloud Computing Platforms
Frederic Desprez
 
Next Generation Grid: Integrating Parallel and Distributed Computing Runtimes...
Geoffrey Fox
 
Opportunities for X-Ray science in future computing architectures
Ian Foster
 
Share and analyze geonomic data at scale by Andy Petrella and Xavier Tordoir
Spark Summit
 
Ad

Viewers also liked (19)

PPTX
Konsep sistem informasi
sitianisai
 
PPTX
A Smarter Order Management System: Why It's Time
B2BandMFT
 
DOC
数据库面试笔试题及答案集
庆顺 程
 
PDF
L10 key slides sunkara
t7260678
 
PDF
How Overseas Driving Licenses Work in Australia
donaldwright54
 
PPT
10 Tips to Modernize Your PTA
Nichole Montoya
 
PDF
Sistemas de informacion hospitalaria HIS
Victor Blanco
 
PPT
Munneasrm2009abstracto6 12564080005062-phpapp03
t7260678
 
PPTX
Writing in creative media
jackgirdlestone1
 
PPTX
Teorias organizativas nicolass
Esteban Farfan
 
PPTX
Pengenalan kepada internet
LayHar
 
PPTX
Wonder miranda ii
20folkerth_j
 
PPTX
Wonder Miranda
20folkerth_j
 
PPTX
Share Success With Others To Get More Visitors!
Keith Jones
 
PPTX
Reka bentuk dan model pangkalan data
LayHar
 
PPT
Λίμερικ
Μπέττυ Κυρίτση
 
PPTX
Wonder character power point
20steiner_j
 
PDF
Sweet spot
t7260678
 
PPT
What is air conditioning repairs and how it is done
www.cowanair.com.au
 
Konsep sistem informasi
sitianisai
 
A Smarter Order Management System: Why It's Time
B2BandMFT
 
数据库面试笔试题及答案集
庆顺 程
 
L10 key slides sunkara
t7260678
 
How Overseas Driving Licenses Work in Australia
donaldwright54
 
10 Tips to Modernize Your PTA
Nichole Montoya
 
Sistemas de informacion hospitalaria HIS
Victor Blanco
 
Munneasrm2009abstracto6 12564080005062-phpapp03
t7260678
 
Writing in creative media
jackgirdlestone1
 
Teorias organizativas nicolass
Esteban Farfan
 
Pengenalan kepada internet
LayHar
 
Wonder miranda ii
20folkerth_j
 
Wonder Miranda
20folkerth_j
 
Share Success With Others To Get More Visitors!
Keith Jones
 
Reka bentuk dan model pangkalan data
LayHar
 
Λίμερικ
Μπέττυ Κυρίτση
 
Wonder character power point
20steiner_j
 
Sweet spot
t7260678
 
What is air conditioning repairs and how it is done
www.cowanair.com.au
 
Ad

Similar to Ph. D. Final Dissertation SLides (20)

PPT
Open Analytics Environment
Ian Foster
 
PPTX
TransPAC3/ACE Measurement & PerfSONAR Update
International Networking at Indiana University
 
PPTX
Enabling semantic integration
Jean-Paul Calbimonte
 
PPTX
Ingredients for Semantic Sensor Networks
Oscar Corcho
 
PPTX
Ontology based top-k query answering over massive, heterogeneous, and dynamic...
Daniele Dell'Aglio
 
PPTX
The Pacific Research Platform
Larry Smarr
 
PPTX
Toward a National Research Platform to Enable Data-Intensive Open-Source Sci...
Larry Smarr
 
PPTX
Semantic Sensor Networks and Linked Stream Data
Oscar Corcho
 
PDF
Reflections on Almost Two Decades of Research into Stream Processing
Kyumars Sheykh Esmaili
 
PPT
A Look into the Apache OODT Ecosystem
Chris Mattmann
 
PPT
An Ad-hoc Smart Gateway Platform for the Web of Things (IEEE iThings 2013 Bes...
Darren Carlson
 
PDF
AIM NIAC PNNL-SA-116502
Mark Greaves
 
PPT
Jewei Hans & Kamber Chapter 8
Houw Liong The
 
PPT
Stream Reasoning: Where we got so far. Oxford 2010.1.18
Emanuele Della Valle
 
PPTX
Machine Learning and Hadoop
Josh Patterson
 
PPTX
TERN Facility Portals - Stuart Phinn
TERN Australia
 
PDF
Triplewave: a step towards RDF Stream Processing on the Web
Daniele Dell'Aglio
 
PPT
Computation and Knowledge
Ian Foster
 
PPTX
Transient and persistent RDF views over relational databases in the context o...
Nikolaos Konstantinou
 
PPTX
How to use NCI's national repository of big spatial data collections
ARDC
 
Open Analytics Environment
Ian Foster
 
TransPAC3/ACE Measurement & PerfSONAR Update
International Networking at Indiana University
 
Enabling semantic integration
Jean-Paul Calbimonte
 
Ingredients for Semantic Sensor Networks
Oscar Corcho
 
Ontology based top-k query answering over massive, heterogeneous, and dynamic...
Daniele Dell'Aglio
 
The Pacific Research Platform
Larry Smarr
 
Toward a National Research Platform to Enable Data-Intensive Open-Source Sci...
Larry Smarr
 
Semantic Sensor Networks and Linked Stream Data
Oscar Corcho
 
Reflections on Almost Two Decades of Research into Stream Processing
Kyumars Sheykh Esmaili
 
A Look into the Apache OODT Ecosystem
Chris Mattmann
 
An Ad-hoc Smart Gateway Platform for the Web of Things (IEEE iThings 2013 Bes...
Darren Carlson
 
AIM NIAC PNNL-SA-116502
Mark Greaves
 
Jewei Hans & Kamber Chapter 8
Houw Liong The
 
Stream Reasoning: Where we got so far. Oxford 2010.1.18
Emanuele Della Valle
 
Machine Learning and Hadoop
Josh Patterson
 
TERN Facility Portals - Stuart Phinn
TERN Australia
 
Triplewave: a step towards RDF Stream Processing on the Web
Daniele Dell'Aglio
 
Computation and Knowledge
Ian Foster
 
Transient and persistent RDF views over relational databases in the context o...
Nikolaos Konstantinou
 
How to use NCI's national repository of big spatial data collections
ARDC
 

Recently uploaded (20)

PPT
First Aid Training Presentation Slides.ppt
WilliamMigoya
 
PDF
6.-propertise of noble gases, uses and isolation in noble gases
ka3095980
 
PPTX
_ISO_Presentation_ISO 9001 and 45001.pptx
nadianisar5
 
PPTX
NORMAN_RESEARCH_PRESENTATION.in education
JustineMaluma
 
PPTX
BIOLOGY TISSUE PPT CLASS 9 PROJECT PUBLIC
iamsrinjandatta
 
PPTX
Anesthesia and it's stage with mnemonic and images
panchalkhushal24
 
PDF
MODULE 3 BASIC SECURITY DUTIES AND ROLES.pdf
MagnificoDionicio
 
PPTX
Research Process - Research Methods course
2357011035
 
PPTX
Lesson-7-Gas. -Exchange_074636.pptx
riesiscar2
 
PPTX
Hydrogel Based delivery Cancer Treatment
sgsayan31
 
PPTX
Phylogeny and disease transmission of Dipteran Fly (ppt).pptx
tasnimsume
 
PPTX
Tour Presentation Educational Activity.pptx
PaulineAngeliqueBere
 
PPTX
Introduction-to-Food-Packaging-and-packaging -materials.pptx
pragyanlahkar16
 
PPTX
ART-APP-REPORT-FINctrwxsg f fuy L-na.pptx
ajp27480
 
DOCX
ENGLISH PROJECT FOR BINOD BIHARI MAHTO KOYLANCHAL UNIVERSITY
NewGautamInternateHu
 
PPTX
FINAL TEST 3C_OCTAVIA RAMADHANI SANTOSO-1.pptx
itsvialalalla
 
PPTX
PurpoaiveCommunication for students 02.pptx
RaizaMarquez
 
PDF
Yusen Logistics Group Sustainability Report 2024.pdf
byrongeorge73
 
PPTX
Tablets And Capsule Preformulation Of Paracetamol
sgsayan31
 
PPTX
MERISTEMATIC TISSUES (MERISTEMS) PPT PUBLIC
iamsrinjandatta
 
First Aid Training Presentation Slides.ppt
WilliamMigoya
 
6.-propertise of noble gases, uses and isolation in noble gases
ka3095980
 
_ISO_Presentation_ISO 9001 and 45001.pptx
nadianisar5
 
NORMAN_RESEARCH_PRESENTATION.in education
JustineMaluma
 
BIOLOGY TISSUE PPT CLASS 9 PROJECT PUBLIC
iamsrinjandatta
 
Anesthesia and it's stage with mnemonic and images
panchalkhushal24
 
MODULE 3 BASIC SECURITY DUTIES AND ROLES.pdf
MagnificoDionicio
 
Research Process - Research Methods course
2357011035
 
Lesson-7-Gas. -Exchange_074636.pptx
riesiscar2
 
Hydrogel Based delivery Cancer Treatment
sgsayan31
 
Phylogeny and disease transmission of Dipteran Fly (ppt).pptx
tasnimsume
 
Tour Presentation Educational Activity.pptx
PaulineAngeliqueBere
 
Introduction-to-Food-Packaging-and-packaging -materials.pptx
pragyanlahkar16
 
ART-APP-REPORT-FINctrwxsg f fuy L-na.pptx
ajp27480
 
ENGLISH PROJECT FOR BINOD BIHARI MAHTO KOYLANCHAL UNIVERSITY
NewGautamInternateHu
 
FINAL TEST 3C_OCTAVIA RAMADHANI SANTOSO-1.pptx
itsvialalalla
 
PurpoaiveCommunication for students 02.pptx
RaizaMarquez
 
Yusen Logistics Group Sustainability Report 2024.pdf
byrongeorge73
 
Tablets And Capsule Preformulation Of Paracetamol
sgsayan31
 
MERISTEMATIC TISSUES (MERISTEMS) PPT PUBLIC
iamsrinjandatta
 

Ph. D. Final Dissertation SLides

  • 1. SuNDroPS: Semantic and dyNamic Data in a Pervasive System Context-ADDICT Revisited Doctoral Dissertation of: Emanuele Panigati Advisor: Prof. Letizia Tanca Co-Advisors: Prof. Fabio A. Schreiber, Prof. G. Cugola Politecnico di Milano { Dipartimento di Elettronica, Informazione e Bioingegneria December 10th, 2014
  • 2. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions & Future Works Summary of the content Introduction and Motivation Go A motivating scenario: the Green Move project Go The SuNDroPS system Go The SuNDroPS legacy blocks Go Context-ADDICT Go PerLa & Tesla Go Hystorical data analysis Go The SuNDroPS new components Go New features of PerLa Go New features of Tesla/TRex: SemTRex Go MR-Miner & MREClaT Go Testing SuNDroPS in the Green Move scenario Go Conclusions and
  • 3. nal remarks Go E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 4. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions & Future Works Introduction and Motivation Users are surrounded by a high quantity of heterogeneous data, often in the form of data streams Humans cannot fully exploit the whole richness of these data without digital support for their analysis Real-time, on-the- y and historical data processing are equally necessary to obtain useful knowledge E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 5. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions & Future Works A Motivating Scenario: the Green Move Project Green Move is a zero-emission vehicle-sharing system which supports the users with additional digital services The Green Move project has been used as a real-world on-
  • 6. eld test, using several SuNDroPS components The user experience is
  • 7. nely personalized based on the user context and on contextual preferences, so the data management process must consider a contextual tailoring of the data Data coming as streams from dierent kinds of sensors (e.g., on-board vehicle status sensor, environmental sensor, . . . ) must be processed on-the- y in order to give to users an immediate feedback and empower their experience. Dierent Information Flow Processing systems have been considered to perform this task (PerLa TRex) E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 8. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis SuNDroPS Architecture Overview SuNDroPS allows to manage the ow of (possibly semantically enriched) information contained in data streams and to study how to extract useful knowledge from it and from the more traditional, static datasets. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 9. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis SuNDroPS Architecture Overview SuNDroPS allows to manage the ow of (possibly semantically enriched) information contained in data streams and to study how to extract useful knowledge from it and from the more traditional, static datasets. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 10. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Legacy: The Context-ADDICT system Allows to query dierent and heterogeneous data sources providing a single entry-point for queries Automatically tailors the query according to the current context of the user, and rewrites it, integrating the results coming from each dierent data source E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 11. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Legacy: PerLa Tesla Two information ow processing systems PerLa is based on a DSMS paradigm Tesla/TRex is based on a Complex Event Processing paradigm E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 12. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis More on Information Flow Processing Two dierent approaches: DSMSs, developed by the database community, consider a data stream as a sequence of tuples, processing them using SQL-like query languages CEPs, developed by the distributed software engineering community, consider the stream as a sequence of events and process them using rule and/or logic based languages for temporal pattern detection PerLa is an example of the
  • 13. rst kind of systems while Tesla/TRex belongs to the second category. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 14. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis More on Information Flow Processing Two dierent approaches: DSMSs, developed by the database community, consider a data stream as a sequence of tuples, processing them using SQL-like query languages CEPs, developed by the distributed software engineering community, consider the stream as a sequence of events and process them using rule and/or logic based languages for temporal pattern detection PerLa is an example of the
  • 15. rst kind of systems while Tesla/TRex belongs to the second category. We next compare the features of the two approaches. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 16. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { PerLa Data VehicleData greenBox id Timestamp Speed A300A 10/12/2014 14:00 0.0 B400B 10/12/2014 14:15 15.0 C100C 10/12/2014 14:01 50.0 TakenOrReleased greenBox id Timestamp takenReleased A300A 10/12/2014 7:00 TAKEN A300A 10/12/2014 8:00 RELEASED B400B 10/12/2014 10:00 TAKEN B400B 10/12/2014 11:00 RELEASED C100C 10/12/2014 12:00 TAKEN E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 17. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { PerLa Queries PerLa Low Level Query CREATE SNAPSHOT MostRecentUse (greenBox id String, takenReleased Integer, date[3] Integer) WITH DURATION 10 AS LOW: EVERY 30 m SELECT greenBox id, takenReleased, date[3] HAVINGdate = MAX(date, 10) UP TO 30m SAMPLING ON EVENT takenInCharge Released PerLa High Level Query CREATE OUTPUT STREAM Theft (greenBox id String, recentUsage date) AS HIGH: EVERY 10 m SELECT greenBox id, MAX(MostRecentUse.date) as mass FROM MostRecentUse, TakenOrReleased, VehicleData WHERE VehicleData.greenBox id = TakenOrReleased.greenBox id AND TakenOrReleased.date = mass AND TakenOrReleased.takenReleased = 0 AND VehicleData.speed 0 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 18. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { PerLa Query Results MostRecentUse greenBox id takenReleased date A300A 10/12/2014 8:00 RELEASED B400B 10/12/2014 11:00 RELEASED C100C 10/12/2014 12:00 TAKEN Theft greenBox id RecentUsage B400B 10/12/2014 11:00 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 19. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { Tesla Data Rule Events event id : 17; greenBox id : A300A; ts : 10=12=2014 14 : 00; speed : 0:0 event id : 17; greenBox id : B400B; ts : 10=12=2014 14 : 15; speed : 15:0 event id : 17; greenBox id : C100C; ts : 10=12=2014 14 : 01; speed : 50:0 event id : 112; greenBox id : A300A; ts : 10=12=2014 7 : 00 event id : 121; greenBox id : A300A; ts : 10=12=2014 8 : 00 event id : 112; greenBox id : B400B; ts : 10=12=2014 10 : 00 event id : 121; greenBox id : B400B; ts : 10=12=2014 11 : 00 event id : 112; greenBox id : C100C; ts : 10=12=2014 12 : 00 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 20. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { Tesla Data Rule Events event id : 17; greenBox id : A300A; ts : 10=12=2014 14 : 00; speed : 0:0 event id : 17; greenBox id : B400B; ts : 10=12=2014 14 : 15; speed : 15:0 event id : 17; greenBox id : C100C; ts : 10=12=2014 14 : 01; speed : 50:0 event id : 112; greenBox id : A300A; ts : 10=12=2014 7 : 00 event id : 121; greenBox id : A300A; ts : 10=12=2014 8 : 00 event id : 112; greenBox id : B400B; ts : 10=12=2014 10 : 00 event id : 121; greenBox id : B400B; ts : 10=12=2014 11 : 00 event id : 112; greenBox id : C100C; ts : 10=12=2014 12 : 00 Rule DEFINE Theft (ID : String) FROM VehicleData (greenBox id = $id AND speed 0) AND LAST Release (greenBox id = $id) WITHIN 10day FROM VehicleData AND NOT Taken (greenBox id = $id) BETWEEN Release AND VehicleData WHERE ID = VehicleData.greenBox id E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 21. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check if a Vehicle is Being Stolen { Tesla Data Rule Events event id : 17; greenBox id : A300A; ts : 10=12=2014 14 : 00; speed : 0:0 event id : 17; greenBox id : B400B; ts : 10=12=2014 14 : 15; speed : 15:0 event id : 17; greenBox id : C100C; ts : 10=12=2014 14 : 01; speed : 50:0 event id : 112; greenBox id : A300A; ts : 10=12=2014 7 : 00 event id : 121; greenBox id : A300A; ts : 10=12=2014 8 : 00 event id : 112; greenBox id : B400B; ts : 10=12=2014 10 : 00 event id : 121; greenBox id : B400B; ts : 10=12=2014 11 : 00 event id : 112; greenBox id : C100C; ts : 10=12=2014 12 : 00 Rule DEFINE Theft (ID : String) FROM VehicleData (greenBox id = $id AND speed 0) AND LAST Release (greenBox id = $id) WITHIN 10day FROM VehicleData AND NOT Taken (greenBox id = $id) BETWEEN Release AND VehicleData WHERE ID = VehicleData.greenBox id Results event id : 301; greenBox id : B400B; ts : 10=12=2014 14 : 15 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 22. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { PerLa Data VehicleData greenBox id GPS Speed A300A 45.1,15.1 30.0 B400B 45.1,20.2 15.0 C100C 42.2,15.1 50.0 Weather GPS Climate Limit 45.1,15.1 Normal 130 45.1,20.2 Rain 90 42.2,15.1 Ice 30 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 23. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { PerLa Queries PerLa Low Level Query CREATE STREAM WeatherChange (position gps data, climate String) AS LOW: EVERY 10 m SELECT position, climate SAMPLING ON EVENT WeatherChanged WHERE climate = Rain OR climate = Ice OR climate = Snow OR climate = Fog REFRESH EVERY 5 m PerLa High Level Query CREATE OUTPUT SNAPSHOT DangerousDriving (greenBox id String) WITH DURATION 2 h AS HIGH: SELECT greenBox id FROM VehicleData, WeatherChange, Weather WHERE VehicleData.gps data = WeatherChange.position AND WeatherChange.climate = Weather.climate AND VehicleData.speed Weather.limits E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 24. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { PerLa Query Results DangerousDriving position climate 45.1,20.2 Rain 42.2,15.1 Ice WeatherChange greenBox id C100C E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 25. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { Tesla Data Events event id : 17; greenBox id : A300A; pos : 45:1; 15:1; speed : 30:0 event id : 17; greenBox id : B400B; pos : 45:1; 20:2; speed : 15:0 event id : 17; greenBox id : C100C; pos : 42:2; 15:1; speed : 50:0 event id : 40; pos : 45:1; 15:1; climate : Normal; temp : 20 event id : 40; pos : 45:1; 20:2; climate : Rain; temp : 17 event id : 40; pos : 42:2; 15:1; climate : Ice; temp : 2 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 26. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Check User Driving Style w.r.t. Weather Conditions { Tesla Rules Results Rule (Rain) DEFINE DangerousDrivingRain(ID : String) FROM VehicleData(speed90) AND LAST Weather(VehicleData.pos-xposVehicleData.pos+x) WITHIN 1h FROM VehicleData AND Weather.climate=rain WHERE DangerousDrivingRain.ID = VehicleData.greenBox id Rule (Ice) DEFINE DangerousDrivingIce(ID : String) FROM VehicleData(speed50) AND LAST Weather(VehicleData.pos-xposVehicleData.pos+x and temp0) WITHIN 1h FROM VehicleData WHERE DangerousDrivingRain.ID = VehicleData.greenBox id Results event id : 340; greenBox id : C100C E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 27. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Legacy: Hystorical Data Analysis Data Mining allows to extract knowledge from the gathered data, discovering previously unknown facts from them. Frequent Itemset Mining
  • 28. nds in the database all the sets of items whose frequency is above a given support threshold Several algorithms are available to perform this task: A Priori Partition FP-Growth EClaT E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 29. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Architecture Overview Legacy: The Context-ADDICT system Legacy: PerLa Tesla Legacy: Hystorical Data Analysis Legacy: Hystorical Data Analysis Data Mining allows to extract knowledge from the gathered data, discovering previously unknown facts from them. Frequent Itemset Mining
  • 30. nds in the database all the sets of items whose frequency is above a given support threshold Several algorithms are available to perform this task: A Priori Partition FP-Growth EClaT E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 31. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT New Components in the Big Data Era SuNDroPS Adds new features to Context-ADDICT: Monitors the environment directly, using sensors, also reasoning on the gathered data Automatically infers (part of) the user context from the environmental data that have been sensed Integrates historical data processing with analysis operations, introducing a new parallel Data Mining algorithms E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 32. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT New features of PerLa The PerLa middleware has been completely reengineered to include asynchronous behaviors of sources (sensors, web services, . . . ) Distributed PerLa allows to exploit the sources (and network) computation power PerLa for Context explicitly integrates the context-aware approach of Context-ADDICT with Context-Oriented Programming COP, allowing sensors to behave dierently according to their current context E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 33. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT New features of Tesla/TRex: SemTRex Original TRex cannot interact with static data SemTRex adds a RDF static data repository to TRex and new operators in the Tesla language (IN) Integrating RDF repositories allows reasoning on the data IN allows to: Enrich the events, including into them facts retrieved from the KB Filter the events using facts included in the KB Prefetching and Caching of data become necessary to keep reasonable response times: basic cache, parametric basic cache, frequent data cache, combined caching strategy. E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 34. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MapReduce-based Frequent Itemset Mining MR-Miner supports the mining processes in SuNDroPS using MREClaT, an EClaT-based algorithm exploiting the MapReduce programming paradigm, that allows SuNDroPS to analyze the data load typical of Big Data scenarios E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 35. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MapReduce-based Frequent Itemset Mining MR-Miner supports the mining processes in SuNDroPS using MREClaT, an EClaT-based algorithm exploiting the MapReduce programming paradigm, that allows SuNDroPS to analyze the data load typical of Big Data scenarios E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 36. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MR-Miner MREClaT { Algorithm Details First step: Mine 1-frequent itemsets E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 37. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MR-Miner MREClaT { Algorithm Details Second step: Mine 2-frequent itemsets E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 38. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT MR-Miner MREClaT { Algorithm Details Third step: Mine k-frequent itemsets E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 39. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT Experiments E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 40. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works The SuNDroPS New Components New features of PerLa New features of Tesla/TRex MapReduce-based Frequent Itemset Mining : MR-Miner MREClaT Pre
  • 41. x Extension Experiments E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 42. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Testing SuNDroPS in the Green Move Scenario PerLa context-aware sensors SemTRex Pervasive and context-aware information push Context-aware vehicle assignment to user reservation, based on contextual user preferences E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 43. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works Conclusions The SuNDroPS system helps users to deal with the high information load that surronds them Context inference based on the environmental sensor data ows Historical data mining using parallel MapReduce algorithms to speed up processing Semantic-enhanced complex event processing using cache to reduce the performance degradation due to the disk bottle-neck Reengineering of the PerLa middleware allowing its distribution on the network components and the integration with other context-oriented paradigms (e.g. Context Oriented Programming) E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 44. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works Future Works Several enhancements are required Complete implementation of Distributed PerLa (currently a prototype) Complete integration of Context Oriented Programming (COP) in PerLa Complete the implementation of caches in SemTRex (currently only the basic and parametric caches are fully implemented) Further testing on the whole system E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 45. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works Thanks E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 46. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works Relevant Publication I 1 A. G. Bianchessi, G. Cugola, S. Formentin, A Morzenti, C. Ongini, E. Panigati, M. Rossi, S. Savaresi, F. Schreiber, L. Tanca, and E. Vannutelli Depoli Green move: A platform for highly con
  • 47. gurable, heterogeneous electric vehicle sharing Intelligent Transportation Systems Magazine, IEEE, 6(3):96{108, Fall 2014 2 E. Panigati Personalized management of semantic, dynamic data in pervasive systems: Context-addict revisited In Proc. of the 2014 International Conference on High Performance Computing Simulation (HPCS 2014), 323-326, 2014 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 48. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works Relevant Publication II 3 F. A. Schreiber, E. Panigati Context-aware software approaches: a comparison and an integration proposal In Proc. of the 22nd Italian Symposium on Advanced database Systems (SEBD), pages 175{184, 2014 4 A. G. Bianchessi, C. Ongini, G. Alli, E. Panigati, S. Savaresi Vehicle-sharing: Technological infrastructure, vehicles, and user-side devices - technological review In Proc. of the 16th International IEEE Conference on Intelligent Transportation Systems - (ITSC), 2013, pages 1599{1604, Oct 2013 5 E. Panigati, A. Rauseo, F. A Schreiber, L. Tanca Pervasive data management in the green move system: a progress report* In Proc. of the 21st Italian Symposium on Advanced Database Systems (SEBD), pages 279{288, 2013 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 49. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works Relevant Publication III 6 E. Panigati, A. Rauseo, F. A Schreiber, L. Tanca Aspects of pervasive information management: an account of the Green Move system In Proc. of the 10th IEEE/IFIP International Conference on Embedded and Ubiquitous Computing, Paphos, Cyprus, Dec 2012 7 G. Alli, L. Baresi, A. G. Bianchessi, G. Cugola, A. Margara, A. Morzenti, C. Ongini, E. Panigati , M. Rossi, S. Rotondi, S. Savaresi, F. A. Schreiber, A. Sivieri, L. Tanca, E. Vannutelli Depoli. Green Move: towards next generation sustainable smartphone-based vehicle sharing In Proc. of SustainIT2012, Oct 2012 E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 50. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works Relevant Publication IV 8 E. Panigati, A. Rauseo, F. A Schreiber, L. Tanca Context-aware information management in the Green Move system { extended abstract In Proc. of the 5th Interop-VLab Workshop, co-Located with ItAIS 2012, Rome, Sept 2012 9 E. Panigati, F. A. Schreiber, C. Zaniolo Data Streams and Data Stream Management Systems Submitted for publication in Data Management in Pervasive Systems - The Shapes and Dynamics of Information in a Pervasive World Book, Springer 10 F. A. Schreiber, E. Panigati Context aware data management and context oriented programming: is convergence possible? Technical Report 2014.7 (DEIB) E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 51. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works Relevant Publication V 11 E. Panigati Methods for Supporting Critical Systems' Failure Diagnosis in the Railway Scenario Technical Report 2013.12 (DEIB) E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System
  • 52. Summary Introduction and Motivation The Green Move Project The SuNDroPS Building Blocks The SuNDroPS System Real-World Testing Conclusions Future Works Conclusions Future Works SemTRex Query Example Send a GM dynamic app to the Green eBox DEFINE SendApp (String: greenBox id, String: appUrl, String:class) FROM VehicleStatus(greenbox id=$a) and ($url, $class) IN (SELECT ?url ?class FROM appKB.rdf WHERE f ?a prop:hasName foo. ?a prop:downloadFromURL ?url. ?a prop:mainClass ?classg) WHERE SendApp.greenBox id=VehicleStatus.greenBox id and SendApp.appUrl = $url and SendApp.class= $class E. Panigati SuNDroPS: Semantic and dyNamic Data in a Pervasive System