Lexicographic Post-processing and Refinement Issues in Wordnets: The Ukrainian Wordnet Experience

Abstract

This paper considers challenges arising within the development of the Ukrainian wordnet, specifically focusing on the integration and alignment of vocabularies originating from diverse sources, such as conventional dictionaries, the Open English Wordnet, and other wordnets. Various cases of synonymy and hyper-hyponymy presented in the Open English Wordnet and in conventional terminological and phonetic dictionaries were analyzed, as well as the definitions of terms in these sources were compared. It was found that traditional dictionaries do not contain all the necessary hyper- and hyponymic relations, and definitions in wordnets are often inaccurate or incomplete. It is shown that creating wordnets based on the "merge" and combined approaches, exemplified by Ukrajinet, necessitates specific lexicographic post-processing. This procedure involves several crucial steps, including incorporating additional lexical units for lacking concepts, refining definitions, and introducing novel concepts and semantic structures, particularly those associated with hyper- and hyponymy, aligned with revised definitions. Keywords: wordnets, Ukrajinet, concept, definition, lexical-semantic relations, synonymy, hyper-hyponymy, lexicographic data