= FCS Taskforce Video Conference 2015-02-11 = * What? * Extension of FCS * Who? * Members of the [[Taskforces/FCS|FCS taskforce]] * When? * 11 February 2015: 10.00 - 11.00 CET * Where? * !FlashMeeting: [http://fm.ea-tel.eu/fm/913371-39355] == Agenda == 0. Welcome 1. Build on SRU 2.0? 2. CQP (w/Corpus Query Language): formal grammar and parser libraries support? 3. POS and universal tags 4. How to proceed == Outcome: Minutes, Decisions and Actions == 1. SRU: * two endpoints needed at least temporarily for SRU 1.2 and 2.0 * provision for version information in the center registry needs to be implemented * 3 versions floating around: also legacy stuff. (old pre-rfc spec) * legacy will be kind of deprecated, but for now remain in support * extended/advanced FCS-spec should be as backwards compatible for RFC-style spec as possible * center registry should contain endpoint version information to make life for the Aggregator easier * '''DECISION''' go for SRU 2.0 2. CQP: * CQP over CQL? no 1-to-1 of CQP Queries over the wire at the centers, will still need some translation (CQP generated by Aggregator to local search engine). but less overhead, complexity, than translating CQP to CQL and back. * Sketch-Engine's flavor of CQP as [[http://corpora.fi.muni.cz/cqp.antlr.txt|ANTLR grammar]] * LREC paper: evaluation and comparison of query languages [http://www.lrec-conf.org/proceedings/lrec2012/summaries/800.html] * '''ACTION''' collect information about query language at endpoint, centers provide information in this [[Taskforces/FCS/Endpoints|table]]. * '''DECISION''' postpone final decision on CQP after there is more data * '''ACTION''' Oliver looks a little more into a FCS-flavor of CQP (others welcome to join) 3. (POS) [[FCS POS tag set|tag sets]]: * universal treebank (new stuff) better as compared to old stuff (universal pos tag set, google) * mostly lossless translation tables for tag sets of the major languages * Link to [[http://quest.ms.mff.cuni.cz/cgi-bin/interset/index.pl interset tagset converter]] * '''DECISION''' general consensus on having a "universal tag set" * maybe (in a distant future) look into translation from specific tag sets to universal in the Aggregator * candidates for "universal thing": universal dependencies, eagles * '''DECISION''' postpone decision on what the "universal tag set" might be * '''ACTION''' Jörg, Pavel S. and Pavel R. look into this and try to come up with a proposal 4. Misc * '''DECISION''' developing and discussing proposals via email should be done on the TF mailing list (= in public) == Documents == * [[https://www.clarin.eu/content/fcs-extension-meeting-minutes|Minutes of FCS Extension Meeting 2014-12-04 in Nijmegen]] * [[http://www.loc.gov/standards/sru/sru-2-0.html|SRU 2.0 specification]] * [[http://cwb.sourceforge.net/files/CQP_Tutorial.pdf|CQP Tutorial]] * [[https://code.google.com/p/universal-pos-tags/|Universal POS]] / [[http://universaldependencies.github.io/docs/|Universal Dependencies]] / [[http://www.ilc.cnr.it/EAGLES96/pub/eagles/corpora/annotate.ps.gz|EAGLES Recommendations for the morphosyntactic annotation of corpora]]