Query Disambiguation Based on Novelty and Similarity User's Feedback

In this paper we propose a query disambiguation mechanism for query context focalization in a meta-search environment. Our methods start from a set of documents retrieved executing a query over a search engine and applies clustering in order to generate distinct homogeneous groups. Then, the following step is to compute for each cluster a disambiguated query that highlights its main contents. The disambiguated queries are suggestions for possible new focalized searches. The ranking of the clusters from which the queries are derived is provided based on a balance of the novelty of cluster contents, and their overall similarity with respect to the query.