2019Q3 Reports: SIGSLAV

From Admin Wiki
Jump to navigation Jump to search

The purpose of SIGSLAV — the ACL Special Interest Group on Slavic Natural Language Processing — established in 2014, is to stimulate research in NLP and creation of linguistic resources for the Slavic languages. The SIGSLAV web page is located at http://sigslav.cs.helsinki.fi. SIGSLAV has 130 members, who are notified about the activities of the SIG through its mailing list sigslav@googlegroups.com.

Officers

The last SIGSLAV elections were held in October 2017 and the following officers were elected by the SIGSLAV members to serve on the SIGSLAV board:

  • Chair: Tomaž Erjavec (tomaz.erjavec@ijs.si), Jožef Stefan Institute, Ljubljana, Slovenia
  • Vice-Chair: ​ Roman Yangarber (roman.yangarber@cs.helsinki.fi), University of Helsinki, Finland
  • Secretary: Preslav Nakov (pnakov@qf.org.qa), Qatar Computing Research Institute, HBKU, Qatar
  • Resource Manager: ​Lidia Pivovarova (pivovaro@cs.helsinki.fi), University of Helsinki, Finland
  • Resource Manager: Josef Steinberger (jstein@kiv.zcu.cz), University of West Bohemia, Czech Republic
  • Interim Resource Manager: Michał Marcińczuk (michal.marcinczuk@pwr.edu.pl), Wroclaw University of Science and Technology, Poland.

The Interim Resource Manager, Michał Marcińczuk, was elected by the other officers of SIGLSLAV, also to help with the preparation of the SIGSLAV shared NER task.

The next election will be held in 2020.

The Advisory Board

The members of the Advisory Board are nominated by the SIGSLAV chair and they continue to serve in this position until further notice. The advisory board currently includes:

  • Natalia Grabar (natalia.grabar@univ-lille3.fr) , Université Lille 3, France
  • Cvetana Krstev (cvetana@poincare.matf.bg.ac.rs), University of Belgrade, Serbia
  • Vladislav Kuboň (vk@ufal.mff.cuni.cz), Charles University, Czech Republic
  • Jakub Piskorski (jpiskorski@gmail.com), Polish Academy of Sciences, Warsaw & Joint Research Centre of the European Commission
  • Adam Przepiórkowski (adamp@ipipan.waw.pl), Polish Academy of Sciences, Poland
  • Serge Sharoff (s.sharoff@leeds.ac.uk), University of Leeds, United Kingdom
  • Kiril Simov (kivs@bultreebank.org), Bulgarian Academy of Sciences, Bulgaria
  • Marko Tadić (marko.tadic@ffzg.hr), University of Zagreb, Croatia
  • Jan Šnajder (jan.snajder@fer.hr), University of Zagreb, Croatia
  • Katerina Zdravkova (katerina.zdravkova@finki.ukim.mk), Ss. Cyril and Methodius University Skopje, Republic of Macedonia

As required by the constitution of SIGSLAV, the officers and advisory board members cover a large number of Slavic languages, i.e., Bulgarian, Croatian, Czech, Macedonian, Polish, Russian, Serbian, and Slovenian.

Events and Publications

In keeping with the tradition that began in 2007, in 2019 SIGSLAV is organising the seventh bi-ennial Balto-Slavic Natural Language Processing Workshop (cf. http://bsnlp.cs.helsinki.fi/). The 7th BSNLP is co-located with ACL in Florence, Italy and takes place on August 2, 2019. Ivan Vulić from the University of Cambridge is giving the keynote talk. The workshop received 20 regular paper submissions, 11 of which were accepted for presentation (55% acceptance rate).

In 2019, SIGSLAV also organized its second shared task (as part of the BSNLP workshop, cf. http://bsnlp.cs.helsinki.fi/shared_task.html), on multilingual named entity recognition. We built a new and significantly larger dataset than for the first shared task, organized in 2017. The data allows systems to be evaluated on recognizing mentions of named entities (NEs) in documents, lemmatization of NEs, and cross-lingual linking of NEs. Four Slavic languages were covered (Bulgarian, Czech, Polish and Russian), and five named entity types (persons, organizations, locations, events, and products). Sixteen teams expressed interest in participating in the NE Challenge, of which eight submitted results, where seven teams worked on NE recognition in all four languages. Five teams also submitted system description papers, which are included as a special section of the BSNLP 2019 proceedings. Together, the workshop presentations cover nine Balto-Slavic languages: Bulgarian, Croatian, Czech, Polish, Russian, Slovak, Slovene, Serbian and Ukrainian.

The next BSNLP Workshop will take place in 2021 and a call for organizing it will go out in due time. Based on the success of the 2019 shared task we also plan to organise the third edition of the multilingual named entity recognition shared task in conjunction with BSNLP’2021.

SIGSLAV would also like to suggest that in the process of evaluation of Workshop proposals for ACL conferences, an important criterion should be not to co-locate workshops with similar topics, as each individually gets less publications than they would have otherwise.

SIGSLAV Business Meetings

The last SIGSLAV business meeting was held on April 5, 2017 in Valencia (Spain). The tentrative plan is for the next business meeting to be held at BSNLP’2019, however, at this point it is not yet clear if any of the officers will be attending, so it might have to be postponed till BSNLP’2021.