Restructuring web search results by generating feedback session and clustering pseudo documents

Restructuring web search results is the best solution for ambiguous queries being entered to the search engine. When ambiguous queries are entered to the search engine gives multiple results for same query, so user don´t get specific and accurate information about what they really want, so it becomes difficult for a user to get specific information related to the submitted keyword. For this reason a new criterion is used in which feedback sessions are first generated from user clicked through logs. Using Feedback session a pseudo documents are generated by calculating TF-IDF (Term Frequency Inverse Data Frequency) vectors for each URL in clicked through logs. Then k-means clustering algorithm is applied and these pseudo documents are clustered and user search goals are generated and restructuring is done through user search goal and user gets specific information fast and correctly. Then the performance of each user search goal is calculated by using CAP metric. These metrics shows how correct restructuring is done. On the basis of this metrics we conducted our experiments on 1720 queries from which 100 most ambiguous queries were separately considered. We observed that the results were high in performance of restructure and less risk.