Lexicon management tools for large textual databases: the Lexinet system

Abstract

This paper deals with the problem of lexicon creation and update for large textual databases. The problem is a particu larly difficult one for research fields where the terminology evolves rapidly. The Lexinet system offers a solution to this problem. It provides a set of statistical and combinatorial algorithms to detect significant items in a corpus of documents from any domain. The results reported concern a corpus of 2,380 titles and abstracts from papers on artificial intelligence, extracted from the French national power company's database. A methodology is descnbed to compare existing terminology with the terminology detected by the Lexinet system.