Pregled bibliografske jedinice broj: 1113500
Towards automatic Semantic Role Labeling in Croatian
Towards automatic Semantic Role Labeling in Croatian // 10th International Conference on Natural Language Processing HrTAL2016. Conference Programme. Book of Abstracts / Tadić, Marko ; Bekavac, Božo (ur.).
Dubrovnik: Hrvatsko društvo za jezične tehnologije, 2016. str. 19-19 (predavanje, međunarodna recenzija, sažetak, ostalo)
CROSBI ID: 1113500 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
Towards automatic Semantic Role Labeling in
Croatian
Autori
Farkaš, Daša ; Filko, Matea ; Štefanec, Vanja ; Merkler, Danijela ; Tadić, Marko
Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, ostalo
Izvornik
10th International Conference on Natural Language Processing HrTAL2016. Conference Programme. Book of Abstracts
/ Tadić, Marko ; Bekavac, Božo - Dubrovnik : Hrvatsko društvo za jezične tehnologije, 2016, 19-19
ISBN
978-953-55375-4-0
Skup
10th International Conference on Natural Language Processing HrTAL2016
Mjesto i datum
Dubrovnik, Hrvatska, 16.06.2014. - 18.06.2014
Vrsta sudjelovanja
Predavanje
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
semantic role labelling, Croatian, Croatian SRL tagset, HOBS
Sažetak
In this paper we present initial steps for semantic role labeling in Croatian. The work builds upon previous research in Croatian Dependency Treebank (HOBS, hobs.ffzg.hr), the set of syntactically annotated Croatian sentences used as a training set for the first Croatian dependency parser. We present the Croatian tagset for SRL and the set of manually annotated sentences which will be used as a training set for automatic SRL system for Croatian. We also present the new version of HOBS, extended with sentences used in our e- course for learning Croatian (www.hr4eu.eu). The sentences used in lessons are annotated both on syntactic and semantic level, as a new tool in computer aided language learning (CALL). Our presumption is that language resources are valuable not only in NLP, but also in CALL. Therefore we have also developed the visualizer that provides users with the visual representation of the set of sentences presented here, depending on the grammatical level they are interested in (MSD, syntactic functions, semantic roles).
Izvorni jezik
Engleski
Znanstvena područja
Informacijske i komunikacijske znanosti, Filologija
POVEZANOST RADA
Ustanove:
Filozofski fakultet, Zagreb