A New Approach of Clustering Feedback Sessions for Inferring User Search Goals

Abstract

Internet information is growing every day exponentially. In order to find out the exact required information from this web search engines has become absolutely necessary tool for the web users. It has also become more difficult to provide user the required information. When Different users provide an ambiguous query to a search engine, they might be having different search goals. Therefore, it is required to find and analyze user search goals to improve the performance of a search engine and user experience. By representing the results in cluster we find out different user search goals for a query. It has advantages in improving search engine relevance and user experience. It extends the delivery and quality of internet information services to the end user. It also improves performance of Web server system. Query classification, search result reorganization and session boundary detection are the approaches attempt to find out user search goals. But the mentioned approaches has many limitations. A new approach has been implemented that overcomes the limitations and analyze, discover user search goals using feedback sessions. This approach first takes the user search query. For each single result of the search query pseudo-documents are generated. Using K-means++ clustering algorithm, these pseudo-documents are clustered. Each cluster can be considered as one user search goal. Finally in restructured result is given to the user where each URL is categorized into a cluster centered by the inferred search goals. Then depending upon user click through, results are restructured and represented to the user in order to satisfy the information need.
DOI: 10.17762/ijritcc2321-8169.150714