Internet Search by Active Feedback

Modern search engines with arrays of parallel servers can provide efficient Web information retrieval. However, as the returned results are usually enormous mass, finding target information is still time-consuming. To increase the correctness of returned results, different page ranking methods have been introduced. They waited and passively weighted users’ feedback in page ranking to try to improve the precision of searching. Previous studies on Web information retrieval show that the average error ratios are over 20%. In this paper, we base on a heuristic from the librarian to present an interactive information retrieval technology called SAFE (Search with Active FEedback). SAFE uses the concept of balanced tree to provide users extra keywords for feedback. These keywords can guide users to have a proper feedback in further searching. The proposed method can help information retrieval applications (not only Web but also desktop) to speed their responses. Additionally, the same idea can be extended to assist search engines to balance their workloads among parallel and distributed servers.