Lexterm, an open source tool for lexical extraction

Abstract

This paper describes the development and exploitation of an open-source terminology-extraction tool named Lexterm. Lexterm was developed under the RESTAD project, whose aim was to introduce automatic processes in the translation of documents, especially those generated by the universities involved in the project. This tool was designed to be used not only by academic staff but also by any person interested in getting a raw bilingual glossary from a bulk of specific topic-domain documents. We will describe the tool, how it retrieves a list of term candidates with their translation equivalences from bilingual corpora, and how this list feeds machine translation and computerassisted translation systems used in the generation flow of academic documents. We will also present how statistical methods are applied to get a list of term candidates as accurate as possible