Determining Argument Structure Variants by Numerical Optimization

The paper proposes a representation of arguments and adjuncts in terms of probability value vectors, and presents a method to calculate argument structure solely based on information available in the corpus. The method can be extended to verbs with multiple argument structures, where argument struct...

Teljes leírás

Elmentve itt :
Bibliográfiai részletek
Szerzők: Szécsényi Krisztina
Szécsényi Tibor
Dokumentumtípus: Könyv része
Megjelent: Palacký University Olomouc 2024
Sorozat:Olomouc Modern Language Series
Language use and linguistic structure: Proceedings of the Olomouc Linguistics Colloquium 2023
Tárgyszavak:
mtmt:35350484
Online Access:http://publicatio.bibl.u-szeged.hu/34690
Leíró adatok
Tartalmi kivonat:The paper proposes a representation of arguments and adjuncts in terms of probability value vectors, and presents a method to calculate argument structure solely based on information available in the corpus. The method can be extended to verbs with multiple argument structures, where argument structure variations associated with the same verb need to be identified. The novelty of the approach is that it relies solely on morphological and syntactic information available in the corpus, discarding the need for meaning-based methods of argument structure identification. The predictions of the probability model proposed in the paper are shown to correlate with the results of a manual annotation control with a high degree of accuracy.
Terjedelem/Fizikai jellemzők:14
210-223
ISBN:9788024465081