Knowledge Organisation and Terminology: Application to Cork

This PhD thesis aims to prove the relevance of texts within the conceptual strand of terminological work. Our methodology serves to demonstrate how linguists can infer knowledge information from texts and subsequently systematise it, either through semi-formal or formal representations. We mainly focus on the terminological analysis of specialised corpora resorting to semi-automatic tools for text analysis to systematise lexical-semantic relationships observed in specialised discourse context and subsequent modelling of the underlying conceptual system. The ultimate goal of this methodology is to propose a typology that can help lexicographers to write definitions.

Based on the double dimension of Terminology, we hypothesise that text and logic modelling do not go hand in hand since the latter does not directly relate to the former. We highlight that knowledge and language are crucial for knowledge systematisation, albeit keeping in mind that they pertain to different levels of analysis, for they are not isomorphic.

To meet our goals, we resorted to specialised texts produced within the industry of cork. These texts provide us with a test bed made of knowledge-rich data which enable us to demonstrate our deductive mechanisms employing the Aristotelian formula: X=Y+DC through the linguistic and conceptual analysis of the semi-automatically extracted textual data. To explore the corpus, we resorted to text mining strategies where regular expressions play a central role.

The final goal of this study is to create a terminological resource for the cork industry, where two types of resources interlink, namely the CorkCorpus and the OntoCork. TermCork is a project that stems from the organisation of knowledge in the specialised field of cork. For that purpose, a terminological knowledge database is being developed to feed an e-dictionary. This e-dictionary is designed as a multilingual and multimodal product, where several resources, namely linguistic and conceptual ones are paired. OntoCork is a micro domain-ontology where the concepts are enriched with natural language definitions and complemented with images, either annotated with meta-information or enriched with hyperlinks to additional information, such as a lexicographic resource. This type of e-dictionary embodies what we consider a useful terminological tool in the current digital information society: accounting for its main features, along with an electronic format that can be integrated into the Semantic Web due to its interoperability data format. This aspect emphasises its contribution to reduce ambiguity as much as possible and to increase effective communication between experts of the domain, future experts, and language professionals.