Abstract
This paper presents some advantages of the conceptual vector model for lexical se- mantics. Besides being a robust representation, this model allows the emergence of relations between terms, as relative synonymy and antonymy. We describe the underlying model, as well as the basic functions which allow to define the notion of thematic proximity. The extension of the indexation mechanism of a term extracted from a document of some speciality domain is mainly based on the notion of vector folding and unfolding between vector spaces. Any term defined thanks to other terms which may not belong to the specialty terminology, will impose the union of the vector space generative families, and lead to vector unfoldings toward a larger base. We show how the lexical distribution, the antonymy, and the synonymy, play as struc- ture spotlight and transversally brigde terms across extended vector spaces. Some experiments focus us on the interest of this kind of approach for terminology.