Different strokes for different folks: an analysis of similarity and diversity in Web search
Maurice Coyle, Barry Smyth · European Conference on Artificial Intelligence · 2004
Relying purely on query-page similarity when ranking Web search results limits the scope of the result set to the detriment of search performance. In this paper we propose that introducing diversity into the ranking metric can increase topic coverage without adversely affecting result relevance in the face of vague queries.