A LINGUISTIC STUDY OF MINING TERMS

Abstract

Terminology mining represents a significant advancement in terminology extraction and involves acquiring and structuring potential terms. Our approach to terminology mining is based on linguistic criteria and a combination of computational methods. In this process, we focus on acquiring complex terms, discovering new terms, and structuring the acquired potential terms. Initially, we provide linguistic specifications for French terms and create a typology of base terms and their variations. Emphasis is placed on effectively handling term variations to construct linguistic structure, identify advanced lexicalization, and optimize the representation of potential term occurrences. Subsequently, we discuss the implemented computational methods, including shallow parsing, morphological analysis, learning morphological rules, and lexical statistics. Finally, we introduce the ACABIT system (Automatic Corpus-Based Acquisition of Binary Terms), outlining its architecture, supported languages, and functions for identifying base terms and their variations.