CLARIN and PAROLE a selfish talk

1
CLARIN and PAROLE a selfish talk

• Steven Krauwer
• Utrecht institute of Linguistics
• UiL-OTS

2
Overview

• What I know about PAROLE in half a slide
• CLARIN in two slides
• First selfish outburst back to the LREC2008
  Panel session
• My assigned task here
• Second selfish outburst sitting on an orphan
• Summary and concluding remarks

3
What I know about PAROLE

• is very little
• lexical project
• late 90s
• morph synt info for gt 10 languages
• semantic info added later (SIMPLE)
• but
• was it a typical project for resources creators,
• or has anyone ever used the results?
• has it been maintained / extended?
• would it be worth including in CLARIN?

4
What is CLARIN

• Common Language Resources and Technology
  Infrastructure (http//www.clarin.eu)
• Basic idea
• European federation of digital archives with
  language data and tools (text, speech,
  multimodal, gesture )
• target audience humanities and social sciences
  scholars
• with uniform single sign-on access to the
  archives
• with access to language and speech technology
  tools to retrieve, manipulate, enhance, explore
  and exploit data
• all languages are equally important
• to cover all EU and associated countries

5
Phasing and funding

• 2008-10 Preparatory phase
• funded by the EU (grant 212230, 4.1 M , 33
  consortium partners in 23 countries, plus 123
  other organisations in 32 countries), with (at
  this moment) additional funding from 19 national
  governments (gt 14 M, ranging 50K - 5M)
• 2011-14 Construction phase
• To be funded by the member states (100 M
  needed, 5 M committed by 1 country, more to
  follow)
• 2015- Exploitation phase
• to be jointly funded by national governments
• 2008-2018 Estimated cost ca 200 M

6
My first selfish outburst

• I see at least 5 ways for CLARIN to fail
• Technical problems?
• dont think so
• Lack of enthusiasm from linguists?
• dont think so
• Lack of resources and technologies
• dont think so

7
How could we then fail?1 standards

• Common standards are crucial for an
  infrastructure of this type where (ideally) all
  resources and technologies can work together
• Threat 1 Lack of agreement on standards
• Our strategy
• single standards not always necessary (support
  more than one, provide mappings between them)
• take existing best and safe practice very
  seriously
• broad involvement of the community at large, not
  just the project partners to ensure broad support
• International collaboration
• teaming up with related initiatives (e.g.
  FLaReNet)
• Role for PAROLE? Is P international? Ps Scope?

8
How could we then fail?2 sustainability

• CLARIN started as a bottom-up initiative but
  major public financial efforts are needed to
  build and exploit the infrastructure
• Threat 2 Lack of financial support
• Our strategy/hope
• EU has put CLARIN on their roadmap of essential
  infrastructures for consideration by national
  governments
• We try to persuade national funding agencies to
  participate in CLARIN
• Investment per country/language not excessive
• Dont think PAROLE could help here

9
How could we then fail?3 ignoring content
evolution

• CLARINs original constituency is mostly the
  written language community but much of the
  material relevant for e.g. modern historians or
  sociologists includes other linguistic modalities
• Threat 3 CLARIN misses the multimodal boat and
  will be obsolete before its completion
• Our strategy
• reaching out actively to the other communities
• openness, with a standing invitation to others to
  join
• teaming up with initiatives with a broader scope
  than just textual data
• Is PAROLE in this respect broader than CLARIN?

10
How could we then fail?4 language coverage

• One of our ideological pillars is that all
  languages are equal
• Threat 4 Minor languages will effectively be
  excluded because of poor resources and technology
  coverage (no market gt no )
• Our strategy
• the BLARK concept (Basic Language Resource Kit)
  may help to bring all languages to a certain
  minimal level of coverage
• porting of technologies and expertise between
  languages can save time and effort
• joint transnational actions to fill the gaps
• Role for PAROLE (esp all languages?)

11
How could we then fail?5 take-up

• CLARIN can only be successful if our target
  audience will actually make use of the facilities
  we offer to support and innovate their research
• Threat 5 Failure to reach and convince our
  audience
• Our strategy
• teaming up with other infrastructure initiatives
  targeting related audiences (e.g. sister project
  DARIAH)
• in next phase strong focus on training, education
  and awareness
• start building convincing demonstrators and joint
  pilots to get a better understanding of the
  users needs
• Not sure whether PAROLE could help here

12
My task here

• The suggested subject of your talk would be an
  overall presentation of the new approaches to
  Linguistic engineering that are represented by
  Clarin, and the key issues that could be taken
  into account from the point of view of the
  producers of Linguistic Resources

13
new approaches to linguistic engineering in
CLARIN

• Not directly we want to use what is (and will
  be) there, except maybe responding to specific
  humanities needs (e.g. tools for diachronic
  research or comparing manuscripts)
• More indirectly we might want to start a new
  fashion by focusing on interoperable tools and
  resources, offered as services, and chainable
  into workflows
• Not sure how PAROLE would come in here

14
key issues that could be taken into account by
producers

• Issue number 1 is .. standards!

15
Issue nr 1

• standards standards standards standards standards
  standards standards standards standards!!!!!
  standards standards standards standards!!!!!!
  standards standards standards standards
  standards!!!!! standards!!!!!!!!!!!!!!!!!
  standards!!!!!!!!!!!!!! standards!!!!!!!!!!!
  standards!!!!!!!!!!!

16
Other issues

• Some more
• Licensing and business models, including
• Fair use
• Repurposing of data
• Quality assurance (incl. definition of quality
  and validation)
• Broad language coverage (esp smaller languages)
• Identifying resources that are really innovative
  or that might help crossing discipline boundaries
• Anything here for PAROLE?

17
Second outburst of selfishness

• The BLARK concept exists, but has become pretty
  much an orphan after ELSNET stopped being funded
• It has been taken up in a number of places (Dutch
  language union, NEMLAR/MEDAR, France, Sweden,
  CLARIN, ELDA), but there is no body that feels
  responsible for the concept
• Could PAROLE play a role here (maybe for the
  lexical part)?

18
Summary and concluding remarks

• My ignorance concerning PAROLE has become evident
  and I need to be filled in but not necessarily
  here and now
• For some issues I really dont know whether
  PAROLE would be capable of playing a role
• If PAROLE represents a significant community it
  might want to participate in standards
  discussions
• PAROLE might be an interesting instrument to help
  looking after (lexical aspects of?) the BLARK
• THANKS!