Semantic matching of natural language Web queries

In this paper, we propose a method to automatically rank documents returned by a search engine in the WWW with respect to a query. The process consists in three steps, the first translates the query and document descriptions into description logic terminologies. The second computes a mapping between related elements in the query and each document. This mapping matches concepts in the terminologies based on their names and their definitions. The last step computes the difference between the query (represented as a terminology) and each document (also represented as a terminology) and ranks the documents according this difference. To deal with linguistic information when comparing description logic concepts, we propose a definition of subsumption that takes into account names similarity between concepts occurring in the descriptions being compared. We describe each step of the method and show the intended results on a running example.