Dealing with specialized co-text in text mining

Abstract

The aim of this paper is to organise lexical and conceptual knowledge by analysing a domain-specific corpus. The domain we focus on is the cork industry. Through the analysis of the corpus, we have found that some common verbs in Portuguese, such as “choose” and “separate” acquire a specialised value in the field under study. This was the starting point for the analysis of the terminological collocations where verbs are the core constituents. For the analysis of these verbal terminological collocations, we used natural language processing techniques, building simple and complex CQL structures with RegEx. The outcome of this analysis permits us to introduce a distinction between polylexical terms and terminological collocations. The terminological collocation is a reality of great relevance in specialised discourse, but unlike terms, it is not defined by conceptual criteria, but by morphological and syntactic criteria.