Naslov
From concept definitions to semantic role labeling in specialized knowledge resources

Autori
Brač, Ivana ; Ostroški Anić, Ana

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of the 13th International Conference of the Asian Association for Lexicography / Gürlek, Mehmet ; Naim Çiçekler, Ahmet ; Taşdemir, Yasin - Istanbul : Asos Publisher, 2019, 604-611

ISBN
978-605-7736-04-8

Skup
13th International Conference of the Asian Association for Lexicography (ASIALEX 2019)

Mjesto i datum
Istanbul, Turska, 19.06.2019. - 21.06.2019

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
terminology ; semantic roles ; semantic role labeling ; specialized knowledge

Sažetak
The paper presents the framework for semantic role labeling that is being developed within the project the Dynamicity of Specialized Knowledge Categories. The main goal of the project is the description of the categories of aviation within semantic frames. One of the tasks towards achieving this goal is developing a methodology for the syntactic and semantic analysis of the language for specialized purposes that could be applied to any specialized domain. The paper first outlines the differences between the terminological database Struna and a lexical database of semantic frames of aviation, AirFrame. Methods applied in analyzingterminological definitions and annotated sentences containing aviation terms are then presented. Conceptual information from the definitionsof key aviation concepts in Struna is compared to semantically annotated sentences extracted from the parallel English-Croatian aviation corpus.The corpus compilation and analysis has been done in Sketch Engine, while the annotation is done in the WebAnno tool. Two approaches to semantic role labelling are discussed. The first approach does not have roles specific to one class, e.g. LIRICS (Petukhova & Bunt, 2008) and VerbNet (Kipper et al., 2008) with hierarchically organized semantic roles. The second approach, developed within FrameNet has more than 2000 frame elements or semantic roles that are verb-specific and frame-specific. Advantages of each approach are discussed, and the semantic tagset applied in the analysis is elaborated. The benefits of using semantic role labeling in terminological resources arefinally discussed.

Izvorni jezik
Engleski

Znanstvena područja
Filologija

POVEZANOST RADA