Identifying Metadata-Specific Collocations in Text Corpora

Varování

Publikace nespadá pod Ekonomicko-správní fakultu, ale pod Fakultu informatiky. Oficiální stránka publikace je na webu muni.cz.
Autoři

HERMAN Ondřej JAKUBÍČEK Miloš KOVÁŘ Vojtěch

Rok publikování 2021
Druh Článek ve sborníku
Konference Proceedings of Electronic Lexicography in the 21st Century Conference (7th Biennial Conference on Electronic Lexicography, eLex 2021)
Fakulta / Pracoviště MU

Fakulta informatiky

Citace
Klíčová slova collocations; corpus; meta-data; text types; word sketch
Popis Statistical corpus analysis of collocations is one of the important steps in creating a dictionary entry: collocations may distinguish senses, describe typical phrasemes and idioms and outline the whole picture of a word's behaviour. However, some collocations are domain-specific, typical only in particular contexts, and thus far there has been no easy way to distinguish “general” collocations from those that are predominantly typical in particular domains. In this paper, we present a tool which allows lexicographers to see typical domains in which a particular collocation occurs. We introduce a statistical procedure based on corpus metadata to identify domain-specific collocations in an intuitive way, and we also present a user interface connected to the word sketch feature of the Sketch Engine corpus interface (Kilgarriff et al., 2014a). The new feature can be used in the manual inspection of collocation lists, as well as when using the API or in a semi-automatic post-editing scenario of building a dictionary.
Související projekty:

Používáte starou verzi internetového prohlížeče. Doporučujeme aktualizovat Váš prohlížeč na nejnovější verzi.