Research On Why Questions Of Spatial Keyword Top-k Queries

With the popularization of mobile communication devices and the rapid development of location technology,more and more web objects have geographic location information and descriptive texts.People can use the related software based on location services to retrieve the places of interest around them.As an important query type supporting location-based services,spatial keyword Top-k query retrieves k objects most relevant to the query requirements according to a scoring formula considering both spatial distance and textual similarity.Since different users often tend to have different preferences for query results,after initiating a spatial keyword Top-k query and obtaining the query result,the user may accidentally discover that the result set contains unexpected objects(hereinafter referred to as why objects),wonder why these objects appear in the query result set,and wonder how to modify the query parameters to exclude them from the result set.To explain the reason to the user why these objects appear in the query result set and how to make these why objects excluded from the query result set are called the why questions.If these questions are not answered,the users may doubt the query results,which will reduce the credibility of the entire query model,and further seriously affect the usability of the query results.In order to improve the usability of query results for spatial keyword Top-k queries,this dissertation studies the why questions of spatial keyword Top-k queries.The method of query modification is adopted to solve the why questions of spatial keyword Top-k queries by modifying query keywords and modifying query preferences.In the scheme of modifying query keywords,a scoring formula that comprehensively considers spatial distance and textual similarity is first given to measure the correlation between spatial-textual objects and query requirements.Next,all why objects are excluded from the new query result set by modifying the query keywords and k value of the original query.Since many refined queries can achieve this purpose,a cost function is defined to measure the modification cost of the refined query relative to the original query.Then,in order to organize spatial-textual objects efficiently,an index structure called WQIR-tree is designed.Based on the designed index structure,a corresponding query processing algorithm is proposed,which enumerates the keyword set by increasing the edit distance,and combines the query processing early termination strategy to speed up the process of finding the refined query with the least cost.Finally,a series of comparative experiments are conducted on the two datasets to verify the efficiency of the proposed scheme.In the scheme of modifying query preference,a cost function is also defined to measure the modification cost.Secondly,Since the sample size of different query preferences is too large,the sampling method is used to calculate the sample size required to obtain at least one best-T% refined query through the sampling model,so as to improve the query efficiency at the expense of the quality of the refined query.Then,the spatial keyword Top-k queries are executed for the refined queries corresponding to the preference sample,and the two strategies of stopping the Top-k query early and skipping the Top-k query step are combined to speed up the process of finding the best-T% refined query with the least cost.Finally,comparative experiments are carried out on the two datasets to verify the efficiency of the proposed method.