In order to improve the usability of a search engines, Query Suggestion, a technique for generating alternative queries to Web users, has become an indispensable feature for such systems. By measuring the similarity between queries in the Euclidean space, however, most existing works mainly focus on suggesting relevant queries to the original query while ignoring diversity in the suggestions, which will potentially dissatisfy Web users' information needs. In fact, it is more natural and reasonable to assume that the query space is a sparse manifold. In this paper, we present a novel query suggestion method based on sparse query manifold learning and sparse manifold ranking with sink regions. By turning selected queries and their sparse neighbors into sink regions on the sparse query manifold, our approach can extract query suggestions by simultaneously considering both diversity and relevance in a unified way. Empirical experimental results on a large scale query log show that our approach is able to effectively generate highly diverse as well as semantically related suggestions.
展开▼
