From Terms to Concepts: A Revisited Approach to Local Context Analysis

Pseudo-Relevance Feedback (PRF) is a widely used technique which aims to improve the query representation assuming as relevant the top ranked documents. This should results in better performance as, after the expansion and re-weigh of the original query, the resultant vector should contain all those worth features able to express utterly the user’s information need. This paper presents the application of a pseudo-relevance feedback technique, called Local Context Analysis (LCA), to SENSE (SEmantic N-levels Search Engine). SENSE is an IR system that tries to overcome the limitations of the ranked keyword approach by introducing semantic levels which integrate (and not simply replace) the lexical level represented by keywords. The evaluation shows that this PRF technique is able to work worthily on both the lexical level represented by keywords and the semantic level represented by WordNet synsets.