Ontologie, locuzioni deboli e legami fra entità terminologiche nell’ambito del progetto DIACOM-fr

In this article we present some ongoing work on building the DIACOM-fr lexical network, a resource on the French terminology used in the domain of international trade. The resource aims to describe terminological entities extracted by analysing, with the support of automatic tools, a 10M-token corpus of textual resources from 1850 to 2018. Among the information collected for each terminological entity, the resource includes information on the entity’s semantic type (which concerns specifically multilexemic terminological entities) and the links (if any) to and from different kinds of terminological entities. To formally represent this information, we decided to leverage the languages and technologies of the Semantic Web, experimenting in particular with the usage of two state-of-the-art models for encoding linguistic information, Lemon and Lexfom. The first modelling activities, conducted on some concrete examples of extracted terminological entities, enabled assessing the feasibility of the proposed approach.