Guessing hierarchies and symbols for word meanings through hyperonyms and conceptual vectors

Abstract

The NLP team of LIRMM currently works on lexical disambiguation and thematic text analysis [Lafourcade, 2001]. We built a system, with automated learning capabilities, based on conceptual vectors for meaning representation. Vectors are supposed to encode ideas associated to words or expressions. In the framework of knowledge and lexical meaning representation, we devise some conceptual vectors based strategies to automatically construct hierarchical taxonomies and validate (or invalidate) hyperonymy (or superordinate) relations among terms. Conceptual vectors are used through the thematic distance for decision makingan d link quality assessment.