DBpedia Association hour

Sebastian Hellmann, Dimitris Kontokostas, Julia Holze

http://dbpedia.org

DBpedia Association

• State of the Association
• Technical Goals
• Funding + Vision

​

​

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia Association

• State of the Association
• Technical Goals
• Funding + Vision

​

​

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia Association (non-profit)

• Founded in 2014
• pushed by Sebastian Hellmann
• Operational since January 2016
• Backed up by Institute for Applied Informatics (InfAI), Leipzig
• Support from DBpedia founding members (Soeren, Chris & Kinsley)
• Draft charter online http://wiki.dbpedia.org/dbpedia-association
• All data published by the DBpedia Association should be made available free of charge under a license equivalent to CC-0 or CC-BY without further restriction on commercial use and redistribution.

​

https://pad.okfn.org/p/how-to-improve-DBpedia

https://pad.okfn.org/p/how-to-improve-DBpedia

https://pad.okfn.org/p/how-to-improve-DBpedia

Bootstrapping the Advisory Committee and Community Committee is next on the todo list for the association

Markus Freudenberg

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia Members

​

​

​

​

Organisational members:

• Fraunhofer IAIS, OpenLink (Board Members)
• iMinds, KB, Huygens (Dutch Chapter)
• Linked Open Data Initiative (LODI, Japanese Chapter)

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia Chapters: Dutch DBpedia

Cooperation between:

• Huygens ING, iMinds/Univ.Gent, Vrije Universiteit Amsterdam, Institute for Sound and Vision, Koninklijke Bibliotheek (National Library)
• and the NL-DBpedia community

Manifested by a Memorandum of Understanding (MoU)

​

​

12 sept 2016

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia Chapters: Dutch DBpedia

• Intentions expressed in the MoU:
• support the goals of the DBpedia Association
• strengthen the Dutch DBpedia chapter and its community of contributors and users
• improve the cooperation with Dutch research infrastructure (Digital Humanities, Linguistics) and Dutch Digital Heritage Network
• Take responsibility, on a best effort basis:
• keep NL-DBpedia healthy
• implement/initiate new developments (specific data extraction)
• organize the community: website, local community meetings
• Chapter is executive part of the Association and receives:
• Support for organisation of meetings
• Financial support for hosting/infrastructure (if income available)
• MoU available on the Association website as inspiration for other chapters!

​

https://pad.okfn.org/p/how-to-improve-DBpedia

Board Meetings

• May and August 2016
• Public minutes
• Resolution resulted in a charter change:

8.4 Voluntary Payment Option for membership fee

• Applicants can apply for reduction
• Board approval needed, generally accepted for core contributors
• Fees are reduced to 20€ per year, but member can pay full fee voluntarily

https://pad.okfn.org/p/how-to-improve-DBpedia

PR and Dissemination

• Website
• Major redesign, still some work todo
• Community can publish projects to the project section
• Twitter
• Facebook
• DBpedia Blog
• Flyers
• Stickers
• First booth @SEMANTiCS

https://pad.okfn.org/p/how-to-improve-DBpedia

Next steps (during 2017)

• More official chapters
• Adjusting the MoU for other chapters
• Better community synchronization
• Monthly public telcos
• Instantiate the Advisory and Community Board
• Improve services and secure funding (next slides)
• Approve the charter
• Board agreed that funding is more urgent than structure

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia Association

• State of the Association
• Technical Progress
• Funding + Vision

​

​

https://pad.okfn.org/p/how-to-improve-DBpedia

Some numbers...

• ~7M Hits/Day (3.3M @ 09/15) see here
• Hoteli Maestral one of our very high traffic users
• 38% improvement in schema conformant data (2015-04 -> 2016-04)
• Measured with RDFUnit
• Data cleanup processes
• Mapping validation tool
• Ontology committee
• 2016-04

https://pad.okfn.org/p/how-to-improve-DBpedia

Release streamline

Too many pre/post/during steps

• Step grouping & automating / Shorten time-to-release

Some highlights

• DataID
• Dockerized triple store / Automated Download page generation
• Automated statistics generation
• Validation checkpoints
• Ensuring correct progress
• Data validation filtering

https://pad.okfn.org/p/how-to-improve-DBpedia

Backlinking

https://github.com/dbpedia/links

• DBpedians can push their links to be loaded into the main endpoint
• (not yet published) Maven/Ant build system that updates, validates and packages links

​

Please wait for the announcement and send a GitHub pull request

(Members can get help from the association for linking and submitting)

​

https://pad.okfn.org/p/how-to-improve-DBpedia

Other things we are busy with

• Further Wikidata integration
• Global pagerank
• Mapping alignment (UPM)
• Mapping discovery & provenance
• GSoC
• DBpedia mappings to RML
• Automated mapping extraction & suggestions
• List / Table extractors
• Lookup improvements
• Topic classification
• Event detection

https://pad.okfn.org/p/how-to-improve-DBpedia

Integration of sources

Two ways:

1. fully automated
1. automatic conversion
2. automatic error detection
2. highly-assisted workflows
• semi-automatic suggestions
• test-driven error detection of data errors
• power tools
• push to source

Mappings automation, validation & provenance

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia Association

• Organisational Structure
• Technical Issues
• Funding + Goals

​

​

https://pad.okfn.org/p/how-to-improve-DBpedia

Funding Goals

• 2017 - build-up phase, DBpedia Association has basic funding for
• Core staff
• Some event and PR money
• However:
• no funding for hosting
• no sustainable funding (bound to projects)
• Goals
• Increase quality of hosting and data
• Provide support for community issues, e.g. service downtime, bugs, travel grants, etc.
• Better publicity and exploitation (events, booths, flyers, add use cases to projects)
• Merge community contributions (work by Heiko, Marco, etc. has not been included yet)
• Systematically develop DBpedias public data, software and services

​

https://pad.okfn.org/p/how-to-improve-DBpedia

Funding Strategies

4 main strategies developed during the DBpedia board discussion:

• Public fundraising (donation campaigns)
• Direct fundraising (direct company support)
• Membership fees
• Community/Project fundraising (i.e. H2020, COST, ITN)

​

Help from the community welcome.

https://pad.okfn.org/p/how-to-improve-DBpedia

Strategy 1: Public fundraising

• Via website
• linked data interface or other services (was this service useful for you?)
• No commitment needed of donator (one-of donation)

Current action items:

• Compile value propositions, Kingsley Idehen made a start: https://medium.com/openlink-software-blog/what-is-dbpedia-and-why-is-it-important-d306b5324f90#.x3n8ope2u
• Wait for more chapters to establish for more impact and chance to donate to chapters directly

​

https://pad.okfn.org/p/how-to-improve-DBpedia

Strategy 2: Direct fundraising

• Large companies use DBpedia
• Several programs exist for supporting open-source projects
• Not effective to write to info@google.com

Current action items:

Step 1: Identify potential organisations: IBM, Google, etc.

Step 2: establish contact

• We are looking for contacts of community members to approach the right person in
• Your ideas are welcome

https://pad.okfn.org/p/how-to-improve-DBpedia

Strategy 2: Subscription fees

• DBpedia is used in education by university courses
• Potential to receive subscription fees by universities and university libraries
• Idea: create a DBpedia educational package
• Teachers have to communicate the need to their university and libraries
• Drawback:
• Currently no know-how in the association how to set this up
• Feedback from university teachers on how to position ourselves

https://pad.okfn.org/p/how-to-improve-DBpedia

Strategy 3: Membership fees

• Good sustainability, i.e. yearly income
• Good for building a network and establish relations
• Does not scale well, i.e. linear twice the members is only twice the income
• Not ideal for core community (invested work)

Following a board decision, we added this paragraph “8.4 Voluntary Payment Option” to the charter:

• Membership fees prevent some community members to join
• reasons no funding or not the right kind of funding or think it is wrong to pay on top of your community work
• Please apply to option, board approves
• Membership fees are reduced to 20 Euro yearly

https://pad.okfn.org/p/how-to-improve-DBpedia

Strategy 4: Community / Project funding

• Differs from other strategies, often members receive the funding directly
• EU grants like H2020 or national funding agencies
• Proposal can contain:
• Deliverables that support DBpedia, i.e. hosting a community event or improve data/software
• DBpedia Association as a third party (similar to Europeana) for support, i.e. 50,000 for hosting or additional data in a release or improved extractors
• Allocation of budget for membership fees of the consortium

Current action items:

• Create a website with useful information and text snippets to copy into proposals
• Jens Lehmann volunteered as direct contact for support: jens.lehmann@cs.uni-bonn.de

https://pad.okfn.org/p/how-to-improve-DBpedia

Strategy 4: Community / Project funding

• Initial monthly telcos about funding
• Not ideal, if ideas evolve not all telco participants can join the proposal
• New strategy:
• Reestablish telcos
• Target “Network funding” such as COST action for traveling and events
• Not as effective, but fair and open

We are looking for a driver, who sets up and maintains these telcos.

If you wish to join, please contact us.

https://pad.okfn.org/p/how-to-improve-DBpedia

How to make DBpedia better?

• Uptime
• Quality
• Coverage
• (new) Services
• Next Meetings

https://pad.okfn.org/p/how-to-improve-DBpedia

https://pad.okfn.org/p/how-to-improve-DBpedia

Thanks for the feedback

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia Groups

Ontology Working Group

Communications Group - Facebook, Slack Discussion about communication channels

Internationalisation Group

DataID

(planned) Wikidata Group, Law?

• Ideas for groups? Association will help set up the group

https://pad.okfn.org/p/how-to-improve-DBpedia

Wikidata

• Great source for DBpedia
• Mission is different from DBpedia (Collect core facts)
• Limited to notability

Integration, Enrichment�Quality Control�Increased Usefulness

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia is ...

https://pad.okfn.org/p/how-to-improve-DBpedia

DBpedia is … a very ambitious project

99.3 % uptime�

25 TB downloadable data

one DBpedia Ontology �

decent linkage�

decent data quality

Identifiers based on�Wikipedia/Wikidata

DBpedia Spotlight ���

⇒ 99.99% uptime for all languages and chapters and spotlight, scalable hosting via the cloud

⇒ 500 TB

⇒ many consistent domain sub-ontologies for each domain from cars to gas turbines to amoeba to star systems to literature

⇒ highly inter- and backlinked, DBpedia can serve as an entry point to find exactly the data (or knowledge) you need

⇒ improved testing and quality control

⇒ Identifiers based on all existing authoritative and robust identifier providers… starting with company data

⇒ all languages, all domains, improved scalability

Funding

• current liquidity is around 10,000€
• some pre-financing via H2020 and German national funding

Services are stable, but lot’s of “what if’s”

• what if OpenLink stops hosting the main endpoint
• what opportunities do we miss without a properly oiled flagship

We need organisations to join and provide financial support, links and backlinks and data.

​

https://pad.okfn.org/p/how-to-improve-DBpedia

Funding

Become a member, if …

• you profit from DBpedia in any way
• you wish to get more in touch with DBpedia
• the services mentioned on the flyer seem useful for your organisation

Include DBpedia in your proposals:

• deliverables
• subcontracting
• consortium member (either association or the national chapters)

https://pad.okfn.org/p/how-to-improve-DBpedia

​

Hope to see you soon.

​

More information: http://dbpedia.org�Sign up: http://dbpedia.org/membership

​

https://pad.okfn.org/p/how-to-improve-DBpedia