Method of Mining Subtopics Using Dependency Structure and Anchor Texts

Abstract

This paper proposes a method that mines subtopics using the co-occurrence of words based on the dependency structure, and anchor texts from web documents in Japanese. We extracted subtopics using simple patterns which reflected the dependency structure, and evaluated subtopics by the proposed score equation. Our method achieved good performance than previous methods which used related or suggested queries from major web search engines. The results of our method will be useful in various search scenarios, such as query suggestion and result diversification.