Increasing Diversity in Web Search Results

The Web is an open repository characterized by the enor- mous quantity and variety of the provided information. A major challenge when searching the Web is to detect this diversity, to reveal it to the user, and to make it exploitable. In this paper, we deal with the notion of diversity in the Web, aiming at constructing a diverse subset of results to answer an ambiguous query or a query for which dierent topics or types of results may exist. We view the problem of diversication as an attempt to achieve good coverage of the whole set of the query results. We consider dierent diversi- cation objectives, as well as corresponding approximation algorithms. We also describe a strategy that relies on topic detection and clustering to diversify search results.