Abstract:
This study investigates the role of hierarchical discourse features in coreference resolution within Russian texts. It evaluates the effectiveness of rhetorical parsers in handling coreference across texts of varying genres and lengths. The paper also identifies key characteristics of rhetorical structure annotation corpora that influence the quality of coreference resolution in diverse linguistic contexts.
Keywords:rhetorical structure theory, deep learning, Russian language, coreference resolution.
Fulltext:
PDF file (384 kB)