KhojYantra: an integrated MetaSearch engine with classification, clustering and ranking

Conventional search engines generally return a long ordered list of results which the users are forced to sift for getting relevant documents. It is envisaged that the search results can be greatly improved by integrating techniques, like ranking, classification and clustering. This paper explores how these three techniques can be integrated into search engine technology to achieve this goal. We introduce KhojYantra, an Integrated MetaSearch Engine with Classification, Clustering and Ranking. Some of the prominent features of the KhojYantra include a better classification scheme to divide the retrieved documents into most relevant, relevant and general groups, its way of clustering the documents and its technique of prioritizing the documents and the clusters according to their relevancy. KhojYantra also provides an easy to browse interface which along with the above techniques enhances the retrieval effectiveness of traditional ranked list. We designed the system as a secondary search engine to reduce the resources necessary to build KhojYantra and to get a larger coverage of the Web. We evaluate the performance of KhojYantra by using the measures, Percentage of Links Followed in the Group, Average Distance Traversed per Document and Documents Visited per Cluster