Supporting flexible queries for autonomous E-Commerce Web databases

The rapid expansion of the World Wide Web has made a large number of E-Commerce Web databases like used cars, houses databases etc. to become accessible to lay users. These users often have vague or imprecise query intentions when searching the E-Commerce Web databases and may not know how to precisely formulate queries that lead to satisfactory answers. This paper proposes a novel approach of the flexible query and results ranking by using query relaxation mechanism in order to present ranked relevant answers to the users. Based on the user original query, we speculate how much the user cares about each attribute and assign a corresponding weight to it. Then, each tuple in the initial query results is relaxed according to the attribute weight and is acted as a relaxed query to obtain the similar tuples from the database. The relaxation order of all attributes and the relaxed degree on each attribute are varied with the attribute weights. For the relevant answers, they are finally ranked according to their similarity to the initial query. The efficiency of our approach is also demonstrated by experimental result.