Linked Open Data and Ontotext Projects

Linked Open Data
Ontologies, Datasets, Projects
CLADA-BG Kickoff, 9 Nov 2018
Vladimir Alexiev, PhD, PMP

Outline
o What is LOD
o ONTO Intro
o ONTO Projects
o CH Ontologies
o CH Projects
o CH Datasets

What is LOD
o Semantic Technologies is a bit of a misnomer
o It's about exposing datasets to machines, not giving a "higher meaning" to data
o Ontologies is a bit of a misnomer
o We're not philosophers and ontologies are not very complex (or, few large datasets use
complex ontologies)
o Ontologies are the data schemas of the semantic web, but not the most important (Shapes
and Application Profiles are now widely used)
o LOD is the essence
o Exposing datasets globally, making each entity/data point addressable (URL)
o "Things not strings"
o Linking them

Where did it come from?
o TimBL
proposal,
CERN,
1989: both
Web and
Semantic
Web
o "Vague but
Exciting"

What does LOD know about TimBL?
o TimBL at
Wikidata
Reasonator
o Names in 50
languages
o Description is
auto-generated
o Parents
confirmed 3
times (with
different details
not shown)

What does LOD know about TimBL?
o Depth of
Information
on TimBL
o Links to ~200
authority files
o Info about ~20
awards
o Life Timeline
o etc, etc

Knowledge Graphs
o Semantically Integrated KB
of a domain, e.g.
o Google KG, e.g. "Jaguar company" vs
"Jaguar cat"
o Springer Nature Science Graph
o Thomson Reuters permid Company Graph
o Microsoft Academic Graph
o Dagstuhl Seminar
Knowledge Graphs: New Directions for Knowledge Representation on the Seman

Thomson Reuters permid Company Graph
(e.g. Sirma Group)

Microsoft Academic Graph (e.g. Ontotext)

Wikidata: EHRI Camps and Ghettos

Wikidata: 98k paintings with image

Wikidata: Number of Creative Works and
Cultural Institutions (SQID)

ONTO History and Essential Facts
o Semantic Web pioneer
o Started in 2000 as a research lab, spun-off and took VC investment in 2008
o 65 staff: 7 PhD, 30 MS, 20 BS, 6 university lecturers
o Over 400 person-years invested in R&D
o Part of Sirma Group Holding: largest Bulgarian software house
o Public company: BSE:SKK, part of SOFIX
o ONTO is core part of Sirma Strategy 2022 with focus on cognitive computing
o Member of multiple industry bodies
o W3C, EDMC, ODI, LDBC, STI, DBPedia Foundation

ONTO Innovation Awards
o Innovative Enterprise of the Year 2017
o EU Innovation Radar Prize 2016 nomination
o BAIT Business Innovation Award 2014
o Innovative Enterprise of the Year 2014
o Washington Post “Destination Innovation” Competition 2014 Award
o Pythagoras Award 2010 for most successful company in EU FP6 projects

Outline
o ONTO Intro
o ONTO Projects
o CH Ontologies
o CH Projects
o CH Datasets

ONTO Innovation (R&D) Projects
• Innovation and Consulting Unit
• More EU research projects than some BG universities
combined
• Consulting projects for banks, cultural heritage institutions,
government institutions, pharmaceuticals
• Focus: semantic data integration, text extraction
• Vertical domains
• Cultural heritage (Europeana Creative, Food and Drink,
EHRI2)
• Companies (EBG, CIMA), innovation (TRR, InnoRate), real
estate data (PDM), agriculture (BigDataGrapes)
• Media/Publishing (TrendMiner, Multisensor, Evala)
• Fact & rumour checking (Pheme, WeVerify)
• Life Science (LarKC, KHRESMOI, KConnect)

Great Variety of Application Domains
EHRI2 European Holocaust Research Infrastructure: transform archival research on the
Holocaust
Evala Cognitive And Semantic Links Analysis and Media Evaluation Platform
euBusinessGraph Enabling the European Business Graph for Innovative Data Products and Services
COMPACT From Research to Policy through Raising Awareness of the State of The Art on Social
Media and Convergence
BigDataGrapes Big Data to Enable Global Disruption of the Grapevine-powered Industries
CIMA Intelligent Matching and Linking of Company Data
Cleopatra Maria Sklodowska-Curie Action: Initial Training Network: Cross-lingual Event-centric
Open Analytics Research Academy
TRR Tracking of FP7 Research Results
WeVerify Factchecking against false news
ExaMode EXtreme-scale Analytics via Multimodal Ontology Discovery & Enhancement
InnoRate Data-driven tools for supporting and improving the decision-making processes of
investors for financing innovative SMEs

ONTO Cultural Heritage Projects

Linked Open Data and Ontotext Projects

↗ User-friendly DB admin and querying
• GraphDB Workbench
↗ REST API for database access
↗ Plugin / Connectors
• GraphDB Engine
GraphDB Semantic Database

OntoRefine: Uplift Tabular Data to LOD
o Easily clean and
import tabular data
o View as RDF in real-time
with virtual SPARQL
endpoint
o Transform using JS & SPIN
o Import newly created RDF
directly to GraphDB
o Usage
o Financial data
o Agricultural data
o CH data, etc

Outline
o ONTO Intro, Products, Clients
o ONTO Projects
o CH Ontologies
o CH Projects
o CH Datasets

o W3C OA Specifications
o Web Annotation Data Model: description
of ontology, use cases and combinations
o Web Annotation Protocol: defines the
interaction between annotation servers
and annotation clients
o Selectors and States: how to select part of
a resource (e.g. section of HTML
document, rectangle from a PNG image,
structural part of a SVG image, page of a
PDF) or specify a particular version of a
resource as it existed at a certain time.
o Embedding Web Annotations in HTML.
Web Annotation o Implementations
o Annotorious image and text annotator by
Austrian Institute of Technology, developed as part of
EuropeanaConnect
o Lorestore server and Annotator OA client by
University of Queensland, Australia
o OACVideoAnnotator by UMD MITH and
Alexander Street Press
o LombardPress annotator of ancient manuscripts that
works over canonic text representations in the
Scholastic Commentaries and Texts Archive
o Annotopia by MIND Informatics group, Massachusetts
General Hospital
o Hypothes.is, largest OA project and development
community. Implements the core AnnotatorJS project. A
number of tools, plug-ins and integrations are available,
including Drupal, WordPress and Omeka integrations.
Omeka is a popular light-weight CMS and virtual
exhibition system
o MangoServer
o Wellcome Quilt, funded by the Wellcome Trust
o Europeana Annotation Server
o Mirador client, a well-known IIIF viewer
o etc etc

OA Example: Bookmarking and Semantic Tagging (Life Science)

OA Example: Annotating SVG Part of Image (ResearchSpace)

o iiif.io Specifications
o Image: semantic description of images
(available resolutions, features, credit line,
conformance level, etc) and serving features
(zooming, gray-scaling, cropping, rotation,
etc)
o Presentation (Shared Canvas): laying
images side by side, assembling folios and
books (using so-called IIIF Manifests), image
annotation. Very popular for virtual
reconstruction of manuscripts, book
viewers, etc
o Authentication: modes or interaction
patterns for getting access to protected
resources (e.g. Login, Click-through, Kiosk,
External authentication)
o Search: search of full-text embedded or
related to image resources (e.g. OCRed or
manually annotated text of some old book)
International Image Interoperability Framework (IIIF)
o IIIF Client Implementations
o Diva.js, especially suited for use in archival book digitization initiatives
o IIPMooViewer, for image streaming and zooming
o Mirador, implementing a workspace that enables comparison of multiple
images from multiple repositories, widely used for manuscripts
o OpenSeadragon, enabling smooth deep zoom and pan
o Leaflet-IIIF, a plugin for the Leaflet framework that also includes display of
geographic maps
o Universal Viewer, widely used by CH institutions
o IIIF Server Implementations
o Cantaloupe, enabling on-demand generation of image derivatives
o IIPImage Server, fast C++ server also used for scientific imagery such as
multispectral or hyperspectral images
o Loris, a server written in Python
o ContentDM, a full-featured digital collection management (DAM) system
o Djatoka, a Java-based image server
o Digilib, another Java-based image server

IIIF Example: Mirador at Biblissima (French manuscript library)

IIIF Example: Search IIIF Images on Europeana

o CIDOC CRM
o Pros: strong foundational ontology, used by
numerous projects especially in Europe.
o Cons: many consider it complicated, some
shortcomings for describing relations
between people and between objects, not
friendly for integrating with other
ontologies, the community (SIG) is slow to
adopt practically important issues, few
application profiles for specific kinds of
objects (e.g. coins vs paintings).
o linked.art
o Pros: a simplified CRM profile created under
the moniker "Linked Open Usable Data
(LOUD)", more developer friendly through
an emphasis on JSONLD, used by some
projects especially in the US.
o Cons: various simplifications that are not
vetted by the CRM SIG, rift with European
CRM developments.
Most Relevant Museum Ontologies
o Schema.org
o Pros: supported by the major search engines thus
ensures semantic SEO and findability, used by the
largest amount of LOD (on billions of websites),
pragmatic and collaborative process for data
modeling with a lot of examples, possible
extensions as exemplified by bibliographic
(SchemaBibEx) and archival extension.
o Cons: not yet proven it is sufficient to represent
rich museum data
o Wikidata
o Pros: universal platform for data integration,
richer model than RDF (but also exposed as RDF),
pragmatic and versatile collaborative process for
data modeling (property creation) with a lot of
examples and justifications, used by some GLAMs
and crowd-sourced projects (e.g. Authority
Control, Sum of All Paintings, Wiki Loves
Monuments).
o Cons: institutional endorsement is not yet strong
enough, concerns of institutions how they can be
masters of "their own" data.

CIDOC CRM
o Conceptual Reference Model (CRM)
o By ICOM, International Committee for Documentation (CIDOC), CRM SIG
o In development for 17 years (since 1999)
o Standardized as ISO 21127:2006 in 2006, continues to evolve
o Current version: CRM 6.2.1 (Oct 2015), version in progress CRM 6.2.3 (May 2018).
o Foundational ontology for history, archeology and art.
o About 85 classes
o About 285 properties (140 object properties and their inverses, and a few that don’t have
inverses)

CRM Graphical Representation : Index

CRM Graphical: Mark and Inscription Information (part 1)

Apply CRM: Model Coins
o E22_Man-Made_Object
o standardized P2_has_type (e.g. Coin from AAT or more specific from Nomisma)
o P56_bears_feature E25_Man-Made_Feature
o P43_has_dimension E54_Dimension with P2_has_type (e.g. die axis), P91_has_unit (e.g. "o'clock"), P90_has_value
o E25_Man-Made_Feature
o standardized P2_has_type: Obverse or Reverse
o P65_carries_visual_item E38_Image (e.g. of a ruler) and/or E34_Inscription (text)
o E38_Image
o P138_represents (e.g. some ruler from ULAN, or e.g. "laurel wreath" from AAT)
o E34_Inscription
o P3_has_note "the text"
o and P72_has_language (e.g. Latin from AAT)
o optionally P73_has_translation to another Linguistic Object

CRM Time Spans
CRM property Meaning Latin phrase Meaning
P82a_begin_of_the_begin started after this moment terminus post quem limit after which
P81a_end_of_the_begin started before this moment terminus a quo limit from which
P81b_begin_of_the_end finished after this moment terminus ad quem limit to which
P82b_end_of_the_end finished before this moment terminus ante quem limit before which

CRM Extensions
o FRBRoo: bibliographic information following FRBR principles (Work-Expression-
Manifestation-Item), artistic performances and their recordings
o PRESoo: periodic publications
o DoReMus: music and performances
o CRMdig: digitization processes and provenance metadata
o CRMinf: statements, argumentation, beliefs
o CRMsci: scientific observations
o CRMgeo: spatiotemporal modeling by integrating CRM to GeoSPARQL
o Parthenos Entities: research objects, software, datasets
o CRMeh (English Heritage): archeology
o CRMarchaeo: archeology, excavation, stratigraphy
o CRMba: buildings
o CRMx: proposed extension for museum objects, including simple properties such as
main depiction of an object, preferred title, extent, etc

Outline
o ONTO Intro, Products, Clients
o ONTO Projects
o CH Ontologies
o CH Projects, Datasets

o Started in 2008
o Has aggregated 53M objects at present
o Perhaps 50-70 Europeana-related projects
o Currently supported by Connecting Europe Facility
as a Digital Service Infrastructure
o Uses Europeana Data Model (EDM), an RDF
ontology
o General search and display mechanism
o The search is not semantic (e.g. won't catch
different multilingual names, unless they are
included in enriched object data)
o A set of fixed facets (including image
characteristics).
Europeana
o Europe (and beyond) GLAM Networking
o Foundation: does the work, ~50 staff
o Association: elections, 2066 members, 75
countries, 19 from BG
o Members Council (36, growing to 50): sets
strategy
o Task Forces: tech guidelines, temporary
o Work Groups: tech guidelines, more
permanent
o Data Quality Council: reflects new strategy

Europeana: Search Paintings of Cupid

Europeana Collections: a "Personal Face"

Europeana Food and Drink: ONTO sem app

Europeana Labs: Galleries of Apps and Datasets

Europeana Data Access: API, OAI PMH, SPARQL

EDM: Typical Graph (from ONTO SPARQL
endpoint)

o Project
o Started 2009, ongoing
o Funded by Mellon Foundation
o Led by the British Museum
o Followed by Yale Center for British Art
(YCBA) and Smithsonian American Art
Museum (SAAM)
o Initial implementation: ONTO, System
Simulation
o Current implementation: Metaphacts
o Additional Involvement: FORTH, Delving
British Museum ResearchSpace
o VRE for Art Research
o CIDOC CRM representation
o Powerful semantic search, saved searches
o Image annotation
o Data basket
o Argumentation
o Intends to be a generic art research
system that can be adapted for various
needs and projects

ResearchSpace: Map British Museum Data to CIDOC CRM

Fundamental Relation "Thing From Place"
CRM Semantic Search

CRM Search: Hierarchical Query Expansion

ConservationSpace: Core System for
Conservationistso Project
o Mellon Funding
o Led by US National Gallery of Art
o Implemented by Sirma Enterprise
o Supported by ONTO
o Uses GraphDB
o Based on Sirma Enterprise Platform
o Sirma MuseumSpace: curation/collection management,
exhibition and loan management, conservation management...
o Semantic integration, enrichment and publication of CH data
o Digital Asset Management
o Thesaurus Management
o Paper-less office (Sirma GO Digital)
o Contract management
o ISO 9001 QMS document management

o Project
o 2-year project (Oct 2015-Nov 2017)
o Mellon funded
o 14 US museums and galleries
o Publish their data to RDF
o ONTO consulted on semantic mapping and data publishing
o Worked alongside two Getty staff (semantic architect and
data architect)
o Publications
o Lessons Learned in Building Linked Data for the American
Art Collaborative, C.Knoblock et al, ISWC 2017: project
challenges, volumetrics and semantic conversion
experience
o American Art Collaborative (AAC) Linked Open Data (LOD)
Initiative: Overview and Recommendations for Good
Practices. E. Fink, 2018
American Art Collaborative
o Achievements
o Aggregated artwork data from 14 institutions: 233,666
Objects, 28,882 Artists and 20,446 other agents (Related
Parties)
o Made about 15M triples. (For comparison, the British Museum
semantic data comprises 2.5M objects and 960M triples.)
o Used a harmonized data model so the data can be shown
together.
o Harmonized not only data models but also value sets to AAT
o Linked per-institution artists to ULAN
o Raised LOD awareness with the target institutions and a wider
audience and mobilized inter-institutional collaboration.
o Some of the institutions took charge of their transformations
to establish a sustainable LOD publication process.
o Created excellent use cases and UI mockups for browsing and
exploration, e.g. comparing artists by style, material and
genres; artwork timelines, etc.

AAC Target Mapping: Actor Gender

o Created as a post-product of AAC
o Application profile for CRM i.e. a particular way of
using CRM.
o Created out of frustration with the complications
of applying CRM, promoted under the moniker
Linked Open Usable Data (LOUD).
o Uses CIDOC-CRM as the core ontology, giving an
event-based paradigm
o Uses the Getty Vocabularies as core sources of
identity, i.e. specific object types (e.g. painting),
activity types (e.g. book binding, gilding, etching),
title types (e.g. artists vs repository title), etc
o JSON-LD as primary RDF serialization. Being JSON,
it is more developer-friendly than other
serializations.
linked.art
o Large number of examples (model
components). Count per area:
o 42 activity,
o 1 concept,
o 2 group,
o 2 identifier,
o 2 legal,
o 1 name,
o 46 object,
o 12 person,
o 6 place,
o 7 set,
o 11 text,
o 2 value

linked.art Representation of Traveling Exhibition: rdfpuml Diagram

linked.art Traveling Exhibition: JSON-LD (left) vs Turtle (right)

GVP LOD Project
o Timeline
o Art and Architecture Thesaurus (AAT): 2014-02
o Thesaurus of Geographic Names (TGN): 2014-08
o Union List of Artist Names (ULAN): 2015-03
o ONTO Services
o Semantic/ontology development
o Contributed to the ISO 25964 ontology (latest standard on thesauri), provided implementation experience, suggestions and fixes.
o Published on varieties of Broader relations (BTG, BTP, BTI)
o Complete mapping specification, comprehensive documentation
o Helped implement R2RML scripts working off Getty's Oracle database, contribution to Perl implementation (RDB2RDF), R2RML
extension (rrx:languageColumn)
o GraphDB semantic repository, clustered for high-availability
o Semantic application development (customized user interface), technical consulting
o SPARQL 1.1 compliant endpoint, sample queries
o Per-entity export files, explicit/total data dumps
o Semantic dataset description (VOID)
o Help desk / support on twitter and google group (continuing)
o GVP LOD is widely regarded as a good example to be followed by GLAMs
Jan

GVP Specialized Hierarchical Relation Inference

GVP Ordered Guide Term, represented as iso:ThesaurusArray

If you have any questions or suggestions,
please email Vladimir.Alexiev@Ontotext.com
Thank you
for your attention!

Linked Open Data and Ontotext Projects

More Related Content

What's hot (20)

Similar to Linked Open Data and Ontotext Projects (20)

More from Vladimir Alexiev, PhD, PMP (20)

Recently uploaded (20)

Linked Open Data and Ontotext Projects

Editor's Notes