Corpus-based knowledge representation in specialized domains.

The ultimate purpose of this work is to describe how the creation of terminological resources can be aided by the use of corpora. The compilation of a specialized corpus provides concepts and terms significant for a specific knowledge domain because as it represents actual current language usage as well as information about their frequency and variation in different contexts. This paper illustrates the methodological approach adopted for the creation of a thesaurus in the domain of tourism. We describe qualitative and quantitative criteria established for compiling a bilingual comparable corpus, the steps involved in semi-automatic terminology extraction and in the definition of semantic relationships between concepts in the controlled vocabulary.