Czech Chapter

https://cs.dbpedia.org/

Milan Dojchinovski

… and about 10 others

AKSW/KILT group @ INFAI @ ULEI

Web Intelligence group @ CTU in Prague

Supporters, contributors and consumers

• University of Economics (UEP)
• Czech Technical University in Prague (CTU)
• Brno University of Technology (VUT Brno)
• Charles University, Faculty of Mathematics and Physics
• University of West Bohemia

​

… probably used by more but hard to collect up-to-date info

Background and history

• Initially started in 2012 as part of the Semantic Web course at the Czech Technical University in Prague
• At tutorials the students worked on new mappings
• First release hosted at the Czech Technical University
• Since 2014 hosted and maintained at the University of Economics, Prague
• Again, contributions partly by students in a Linked Data course, and also within a MSc thesis
• Last update: April 2018

3

Maintenance and funding

Usual maintenance tasks:

• Dataset updates
• Dockerization
• Updates of the Virtuoso triple store
• Update of the landing page: https://cs.dbpedia.org/
• Integration of new datasets

Funding:

• indirectly from national and EU projects (LinkedTV, LOD2, OpenBudgets, …)

Big thanks to Vaclav Zeman for the maintenance efforts!

4

Research behind

• Linked Hypernyms dataset
• Enrichment of DBpedia entities with additional types
• Entityclassifier.eu
• NER system for spotting and linking DBpedia entities
• InBeat
• Rule-based recommender
• DBquiz (https://mynarz.net/db-quiz/)
• EasyMiner (http://www.easyminer.eu/)
• Mining association rules
• Number of master thesis around DBpedia on:
• analysis, visualization, use … of DBpedia

5

Research behind (cont.)

• RDFRules
• Mining frequent patterns in RDF datasets

(?b <http://dbpedia.org/ontology/spokenIn> ?a <dbpedia>)

AND

(?a <hasNumberOfPeople> [ 0.0 ; 655017.5 ) <yago>)

=>

(?a <hasOfficialLanguage> ?b <yago>)

​

(?a <hasChild> ?c) AND (?c <hasGender> ?b) => (?a <hasGender> ?b)

support: 2, headCoverage: 0.01, confidence: 0.5, pcaConfidence: 0.66, lift: 1.26

6

OpenData.cz - a related initiative

7

Wrap-up

Regularly maintained but …

… no clear (strong) incentives

… no regular funding

Some possible future directions:

• more DBpedia focused research (new master and PhD students)
• involve more research institutions
• involve industry (projects, prototypes … which use or contribute)
• stronger collaboration with existing initiatives (e.g. OpenData.cz)
• joint COST proposal with other DBpedia Chapters

8

COST Action

Collaborative COST proposal on:

“DBpedia: the European Open Knowledge Graph”

(EU communities, EU languages, EU data centres, EU research)

Goals:

1. Strengthen the collaboration between DBpedia chapters
2. Scale-up: automate integration, retrieval and fusion of facts
3. Introduce direct societal impact

​

Ultimate goal: DBpedia as the most comprehensive freely available knowledge infrastructure

9

Goal #1: Strengthen the collaboration

• More intensive collaboration and shearability of efforts
• Short-term visits, exchange of experts and expertise
• Joint events (i.e. workshops, tutorials) on specific topics (e.g. NLP)

​

• Integrate existing and ongoing research results
• New datasets, more facts
• Better quality

​

• Provide additional support for the minor DBpedia chapters
• More frequent updates, updated mappings, etc.

10

Goal #2: Scale-up

• Automate deployment of new datasets
• Fully automatic retrieval, integration and fusion of datasets
• Rich metadata and provenance information

​

• Deploy new European (and beyond) DBpedia chapters
• Slovakia, Slovenia, Serbia, Macedonia, baltic countries
• … but also provide support for the minor chapters

11

Goal #3: Direct societal impact

Goal: generate valuable and interesting use cases with a societal benefit -> Targeted domain-specific improvement (coverage and quality)

Possible topics related to:

• (Public) Health (diseases, sport, …)
• Social networks and media (fake news, …)
• Disaster events (earthquakes, tsunami, pollution,…)
• Food and agriculture
• Libraries

(collaboration on the topics above between industry and research)

12

COST Action requirements

• > 50% from Inclusiveness Target Countries (ITC):

Albania, Bosnia-Herzegovina, Bulgaria, Cyprus, Czech Republic, Estonia, Croatia, Hungary, Lithuania, Latvia, Luxembourg, Malta, Montenegro, Poland, Portugal, Romania, Slovenia, Slovakia, Macedonia, Republic of Serbia and Turkey.

• Involving early career investigators (< 8 years from PhD)
• Gender balance (50-50)
• 15 pages proposal, one stage evaluation
• Budget: 129K EUR per year (about four years)

13

COST International Cooperation

• ?Argentina?, Australia, Bangladesh, ?Brazil?, Canada, Chile, China, Colombia, Costa Rica, Hong Kong, India, Indonesia, Iraq, Japan, Korea, Mauritius, ?Mexico?, Namibia, New Zealand, Pakistan, Peru, Saudi Arabia, Singapore, South Africa, Sudan, Thailand, United Arab Emirates, United States of America and Uruguay.

14

Next steps

1) Note your interest in the paper (name, affiliation, contact)

2) We will contact you and ask for an input

3) We will prepare one-pager and refine the idea with you

4) Establish the consortium around the proposal and work on it

​

Proposal deadline: November 29, 2018

InfoDay in Vienna: November 6, 2018 (validate the proposal idea)

​

Feel free write us: dojchinovski@infai.org

15

Supporters

16

Indirectly supported by the EC under

several FP7 and H2020 projects.