期刊名称：International Journal of Computer Science and Information Technologies
电子版ISSN：0975-9646
出版年度：2016
卷号：7
期号：1
页码：234-238
出版社：TechScience Publications
摘要：Internet and web search has become anindispensable part of everyone's life. The usefulness of asearch engine depends on the relevance of the result set itgives back. To organize and bring some order to the massiveunstructured search results, search engines should groupsimilar results together. This is a novel framework torestructure the search results by discovering different usersearch goals for a query. So that users with different searchgoals can find what they want conveniently. This is obtainedby clustering the proposed feedback sessions of usersconstructed from user click-through logs and can efficientlyreflect the information needs of users. Pseudo-documents aregenerated to better represent the feedback sessions forclustering. Distributional clustering is employed to cluster thepseudo-documents. Finally the search results are groupedbased on the label obtained through clustering.
关键词：Click-through logs; feedback sessions;pseudo-documents; user search goals; distributional;clustering; precision and recall
