OpenLinguiscsWorkingGroup (OWLG) · OWLG%ac+vi+es%!...
Transcript of OpenLinguiscsWorkingGroup (OWLG) · OWLG%ac+vi+es%!...
Open Linguis+cs Working Group (OWLG)
Chris+an Chiarcos chiarcos@uni-‐frankfurt.de
Open Knowledge Founda+on (OKFN, hCp://okfn.org)
n non-‐profit organiza+on n founded in 2004 n promote open knowledge in all its forms
q e.g., publica+on of government data (UK, US)
n provide infrastructural support for several working groups
OKFN Open Linguis+cs Working Group (OWLG)
n founded in Oct 2010 in Berlin, Germany n open network of individuals interested in
q linguis+c resources and/or q their publica+on under open licenses
n mul+-‐disciplinary q NLP/CL, typology/language documenta+on, SW, …
n infrastructure q mailing list, web site/blog, wiki q hCp://linguis+cs.okfn.org
OWLG goals (hCp://linguis+cs.okfn.org)
1. Promote open data in rela+on to language data 2. Point of reference and support for open linguis+c data 3. Facilitate communica6on between researchers that use,
distribute, or maintain open linguis+c data 4. Mediate between providers and users of technical
infrastructures 5. Build and maintain an index of open linguis6c data sources 6. Assemble best-‐prac6ce guidelines and use cases concerning
crea+ng, using and distribu+ng data 7. Gather informa6on on legal issues
OWLG goals (hCp://linguis+cs.okfn.org)
1. Promote open data in rela+on to language data 2. Point of reference and support for open linguis+c data 3. Facilitate communica6on between researchers that use,
distribute, or maintain open linguis+c data 4. Mediate between providers and users of technical
infrastructures 5. Build and maintain an index of open linguis6c data sources 6. Assemble best-‐prac6ce guidelines and use cases concerning
crea+ng, using and distribu+ng data 7. Gather informa6on on legal issues these aspects are
specifically well developed
OWLG ac+vi+es
n mostly point-‐to-‐point coopera+ons between individual members
n regular telcos/mee+ngs n workshops -‐> building an interdisciplinary community
q collocated with larger events of different communi+es q Linguis+cs Track of the OKCon, June 2011, Berlin, Germany q Linked Data in Linguis+cs -‐> linguis+cs / NLP
n March 2012, Frankfurt/M., Germany -‐> academic linguis+cs n Sep 2013, Pisa, Italy -‐> NLP/seman+cs n May 2014, Reykjavik, Iceland -‐> NLP
q MLODE-‐2012, Sep 2012, Leipzig, Germany -‐> IT q Linked Data in Linguis+c Typology, Sep 2013, Leipzig, Germany
OWLG ac+vi+es
n point-‐to-‐point coopera+ons between individual members
n regular telcos/mee+ngs n workshops -‐> building an interdisciplinary community
q keeping +es with other communi+es & projects q e.g., Cyberling, W3C OntoLex, ACL SIGANN/SIGLEX q e.g., MPI-‐EVA, LOD2, LIDER, QTLeap
n joint publica+ons and presenta+ons n building and maintaining the Linguis+c Linked Open Data (LLOD) [sub-‐]cloud
LLOD cloud
n a collec+on of linguis+c resources q published under open licenses q as linked data q decentralized developed and maintained q meta data at hCp://datahub.io
=> cloud diagram
q developed as a community effort in the context of the Open Linguis+cs Working Group of the Open Knowledge Founda+on
next: LLOD 2011-2014
Building the Cloud: 2011 A sketch from a table napkin
n ini+ally, we maintained a list of open or representa+ve resources q in Jan 2011, we marked possible synergies
n merely a vision q includes non-‐open resources as placeholders for other resources to come
q not physically realized
n a strong metaphor brought to a new community http://nlp2rdf.lod2.eu/OWLG/llod/2011/01/llod.png
Chiarcos, Hellmann & Nordhoff „Linking Linguis+c Resources“ (2012)
n hypothe6cal linking for selected data sets from NLP, SW and typology described in the book
Closing chapter of the LDL-‐2012 companion volume
Draning the Cloud: LREC-‐2012
„dran status“ hand-‐craned, including resources whose RDF conversion and linking was suggested, not yet performed at the +me
http://nlp2rdf.lod2.eu/OWLG/llod/2012/02/llod.png
Building the Cloud: MLODE-‐2012
n Mul+lingual Linked Open Data for Enterprises q goal: build the first instance of the LLOD cloud q workshop & hackathon
n authors were encouraged to provide data n data conversion, metadata update at hCp://datahub.io
n automa+cally generated diagram q Richard Cyganiac‘s converter scripts
http://sabre2012.infai.org/mlode
Building the Cloud: MLODE-‐2012
http://linguistics.okfn.org/resources/llod/
Building the Cloud: 2013+
n MLODE data post-‐proceedings q Special issue of the Seman+c Web Journal q Prepara+on of addi+onal data sets in the process
n e.g., lemonUby (Eckle-‐Kohler et al., accepted)
n Linked Data in Linguis+c Typology, Aug 2013 q addi+onal poten+al datasets
n lexical databases of Austronesian languages n a database of syllable structures
n Intensified community work
Building the Cloud: Sep 2013
n more data sets not fully linked, yet
n new drawing script q by John McCrae
& Chris+an Chiarcos
n manually categorized and colored q GraphML
n more data sets n more rigid criteria q linked &
accessible
n two-‐layered resource taxonomy
n this (<=) version is merely to eliciate feedback q new diagram end of
May 2014
Building the Cloud: May 2014
Recent developments
n finalizing LLOD diagram revision q for LDL-‐2014, May 27th, 2014
n harmonizing linguis+c resource categories q synchroniza+on with MetaShare categories
n adding new resources q relevant LREC „Share your resources“ datasets ?
n subsequently enforce further constraints on LLOD „bubbles“ q open licenses (currently: accessible ~ LOD diagram) q well-‐formedness / meta data check