De vectores a conceptos: nuevas rutas en la representación computacional del lenguaje

Laburpena

This thesis work addresses the creation of conceptual subspaces from distributional language models, proposing new methodologies to improve the computational representation of concepts. The study also proposes the formalization of a linguistic model based on quantum probability, which provides a new insight into the interpretation of semantic information. Furthermore, several studies are carried out to analyze the applicability of this model for measuring cognitive biases in language use. These biases, inherent in human communication, can be better understood and quantified through the proposed model. The research, therefore, focuses on advancing the understanding y analysis of the cognitive and semantic aspects of language, providing new tools and methods that can be of utility in the field of computational linguistics y cognitive sciences