An Information Retrieval Model Based on Automatically Learnt Concept Hierarchies

The paper investigates the application of fuzzy logic based concept summarization and formal concept analysis in automatically building concept hierarchies from a text corpora. The context of a term has been modeled using its syntactic relations with the most frequent verbs, which act as attributes. This context information has been used to produce a concept lattice, which retains the concept hierarchies as well as the membership weights of the objects. The concepts within each hierarchy have been summarized using a fuzzy logic based soft least upper bound approach. An information retrieval model is proposed, which uses fuzzy formal concepts to get the relevance degree between the document and the query. Results for ontology evaluation are shown on two domain ontologies.