A method for the evaluation of the peculiar lexicon significance

Abstract

In this work, a methodology for semi-automatic derivation of knowledge from document collections is proposed. In order to extract relevant information from documents, a process integrating both statistical and lexical approaches is applied. We propose a strategy for the semantic evaluation of the index terms extracted in order to ensure a good correspondence between the information searched for and the information retrieved. Therefore, we propose a system for the peculiar lexicon extraction and assessment. The system can be used for defining an ontological model to be used in the  semantic processing of a corpus  of documents belonging to a  specialist domain.