Automatic thematic categorization of documents using a fuzzy taxonomy and fuzzy hierarchical clustering

In this paper we formally define the problem of automatic detection of thematic categories in a semantically indexed document, and identify the main obstacles to overcome in this process. Furthermore, we explain how detection of thematic categories can be achieved, with the use of a fuzzy quasi-taxonomic relation. Our approach relies on a fuzzy hierarchical clustering algorithm; this algorithm uses a similarity measure that is based on the notion of context.