Using an automatically generated ontology to improve information retrieval

Abstract. In information retrieval, certain methods were developed to extract semantics from documents. These methods aim at providing some answer to requests made by a user, instead of providing documents that may or may not contain the answer. Also, a vast majority of these techniques were conceived to be applied on the entire World Wide Web and not on a particular field of knowledge, like corporative data. It could be interesting to use domain specific ontologies in trying to link a specific query to related documents and thus, to be able to better answer these queries. This paper presents an approach which proposes to use the Text-To-Onto software to automatically create such a domain specific ontology. It also shows how Text-To-Onto can be linked to a conceptual graph (CG) based search engine for Web documents, Sesei, which already uses Connexor, Notio and WordNet. The resulting software is called SeseiOnto and makes it possible to improve the relevance of documents returned to the user, and identifies with more precision than Sesei, which part of the document should contain the answer to the query. This article presents how the interoperability between CG tools allowed such a system to be made possible. We will also show how conceptual graphs contributed greatly to the processing of natural
language in our application.