User feedback based enhancement in web search quality

Of late, there has been a paradigm shift in web searching from the content based searching to the connectivity based or more commonly known as hyperlink based (or simply link based) searching. But, both the content based approach as well as the link based approach are objective ones, which are totally dependent on the effectiveness of their "feature extraction" mechanisms, with no apparent consideration to the preference of the searcher. In this work, a "user satisfaction" guided web search procedure is proposed. We calculate the importance weight of each document viewed by the user based on the feedback vector obtained from his actions. This document weight is then used to update the index database in such a way that the documents being consistently preferred go up the ranking, while the ones being neglected go down. Our simulation results show a steady rise in the satisfaction levels of the modeled users as more and more learning goes into our system. We also propose a couple of novel additions to the web search querying techniques.