Evolution of user dependent model to predict future usability of a search engine

Abstract

Search Engines are always making efforts to better understand their user's need and improve user satisfaction. This research examines the important issue of user dependency (effectively a combination of loyalty and satisfaction) on web search engines, first studying existing dependency and then modeling that dependency. An algorithm developed to find a quantitative value of “user dependency” on Search Engine is presented. Here, the term ‘user dependency’ implies the psychological satisfaction of a user with the search results presented for a search session. It's an indicative measure of user's trust on Search Engine and impacts the user's choice to use the same Search Engine in future. This paper investigates factors that influence a search session and uses a fuzzy based approach to determine the dependency and overall trust the user places on the Search Engine. The proposed algorithm accepts ‘user rating for the search session’ as input and based on the ‘user satisfaction with search’ generates a value for user dependency. The findings have implications for search engines in improving their ranking algorithms based on explicit user feedback on the search experience. The algorithm has been implemented and tested using Visual Basic environment developed for this study. The validity of algorithm and correctness of its result is evaluated using a survey conducted with a sample of users. Results have been validated for accuracy and their conformance to sampled user's satisfaction.