Contextual information in terminological knowledge bases: A multimodal approach

Abstract

Abstract In Lexicology and Terminology, the notion of context often refers to a textual excerpt selected according to syntactic criteria, leaving aside any semantic and pragmatic analysis. For the creation of terminological knowledge bases (TKBs), this notion is too simplistic. Contexts must be studied from an integrated approach, such as Frame Semantics and Frame-Based Terminology. These theoretical and methodological frameworks do not separate the semantic and pragmatic components of language. Meaning is acquired in context, more specifically, within a frame including a semantic and pragmatic background. Within the domain of the environment, we select and manipulate multimodal information to offer two kinds of contexts to the end-user: (1) FrameNet-like contexts, more specifically, sentences showing the different syntactic constructions of the frame elements and the target predicate; (2) combined contexts, including knowledge-rich linguistic contexts coupled with knowledge-rich visual contexts, which provide a comprehensive view of related processes and specialized lexical units. In the TKB EcoLexicon, the resulting multimodal contexts are structured in terms of specific frames and general events. Thus, the end-users have the possibility to find both cognitive and communicative information, which is selected according to the user's level of expertise.