Build an ontology is a tedious task, which still requires a great amount of manual work. Texts, as knowledge sources, can help, but TALN tools stop at linguistic level. Manual conceptualization fill the gap between a linguistic model and a conceptual model. In this thesis we study how a symbolic clustering method, Formal Concept Analysis, can be combined with a linguistic model to help the knowledge engineer. We have experimented on three different domains represented by same-sized corpora. We show that ontology learning from texts cannot be fully automatized. We propose solutions that combine FCA and terminological analysis, to let the computer suggests useful clusters and faithful representation of texts.
Mendeley saves you time finding and organizing research
There are no full text links
Choose a citation style from the tabs below