Abstract

REYES-MAGANA, Jorge; SIERRA, Gerardo; BEL-ENGUIX, Gemma  and  GOMEZ-ADORNO, Helena. LexAN: Lexical Association Networks. Comp. y Sist. [online]. 2023, vol.27, n.4, pp.955-963.  Epub May 17, 2024. ISSN 2007-9737.  https://doi.org/10.13053/cys-27-4-4774.

This paper presents Lexical Association Networks (LexAN), which entail the development of a mathematical model comprising a collection of words derived from a textual corpus. The interconnections between word tokens are represented by weighted edges within a non-directed graph structure. The construction process of LexAN involves 6 stages: 1) Lemmatization 2) Multi-word expressions 3) Stopwords removal 4) Co-ocurrence graph 5) Word Co-ocurrence norms, and 6) LexAN construction. We employed a Medical text corpus containing 574,011 words to build our graphs. To assess the efficacy of our LexAN, these graph structures were implemented within a tool designed to address the lexical access problem, specifically functioning as a reverse dictionary. This application resulted in favorable and promising results.

Keywords : Network; co-occurrence; lexical access.