A Scoring Function for Retrieving Photo Sets with Broad Topic Coverage

As a storage-unit of user created Web objects, the set has become an emerging challenge in the retrieval. Set search requires relevant sets to meet the information need of users, whereas traditional information retrieval focuses on finding relevant Web objects. This paper proposes a new approach to measure relevance of sets with respect to a user query by their topic coverage. The main idea of the proposed approach is to prefer the set which covers as many different query-related topics as possible. The problem domain of this paper is photo sets of the Flickr.com, which are collections of photos. The experimental result shows that our algorithm outperforms the previous collection selection algorithm.