Abstract: Based on XML twig query fragments relaxation, this paper proposed an approximate querying and results ranking approach to achieve the approximate query results against XML documents; our method gathers the query history to speculate the user's preferences, which is used to calculate the importance for each query fragment of the twig query,and relax the original query according to the sectuence of the fragments’importance; based on the number of query fragments we adopt different relax way; if the number>2，relax the original query according to the granularity of the fragment; if the number<2,rclax the original query according to the granularity of query node,and adopt a different way to relax the numerical query and non-numerical query, and then obtain the most relevant query results. Finally, the relevantctuery results are ranked based on their satisfaction degree to the original query and the user preferences. Our experiment shows that the approximate ctuerying and the results ranking approach can efficiently meet the user's needs and user's preferences, has the high recall and precision.

Key words: Twig query,Approximate query,Query relaxation,User preferences,Results ranking