Refined and diversified query suggestion with latent semantic personalization

The World Wide Web has grown up rapidly and beyond doubt, majority of it contains useful information. But a portion of the web has a junk data and its presence cannot be ignored. Therefore, it becomes a challenging task to retrieval systems to provide useful information to the user. Also, choice of the query plays a crucial role in retrieving documents. In this paper, we propose a model that provides a better query suggestion to the user. Current state of the art models provide suggestions by taking only the relevancy of the query into consideration. Unlike these approaches, this paper, proposes a model that takes both personalized and diversified results into consideration. We propose query refinements to the query and using these refinements, the diversification of the refined query is performed. Consequently, the latent semantic personalization of the input query is done using the user's query log so as to provide dynamic results to the user. Our claims are supported with the experimental results.