Topic-driven web search result organization by leveraging wikipedia semantic knowledge

Effective organization of web search results can greatly improve the utility of search engine and enhance the quality of search results. However, the organization of search results is difficult because the sub-topics of a query are usually not explicitly given. In this paper, we propose a novel topic-driven search result organization method, which can first detect the sub-topics of a query by finding the coherent Wikipedia concept groups from its search results; then organize these results using a topic-driven clustering algorithm; in the end we score and rank the topics using the support vector regression model. Empirical results show that our method can achieve competitive performance.