Research On Technique Of Reverse K-Ranks Query Processing

With the wide application of database technology in the fields of commerce and military,it is increasingly important to analyze and extract massive data.As one of the most important queries in the database field,the attribute-based query can filter out the most interesting query objects for users according to their individual needs.At present,preference queries have bright application prospects in multi-objective decision-making and recommendation systems.Reverse k-Ranks Query,as a typical kind of preference query that has emerged in recent years,can help merchants to find potential consumer groups that are most interested in their products,and then serve business decisions and marketing.This kind of query has advantages that are unmatched by other queries in terms of result set size and query perspective.However,there are still some shortcomings in query diversity and result availability.In view of this,we combine the existing Reverse k-Ranks Query technology with the real-life application scenarios,and study this query from two aspects mentioned above.The main contributions of this thesis are summarized as follows:(1)As existing methods cannot solve the multi-object Reverse k-Ranks Query effectively,we propose a group-based algorithm called GP-RKR.Firstly,a clustering algorithm is used to cluster the query point set according to the similarity of the attribute.Then we propose a layered grid index structure LG-Index to index the data points.In addition,the boundary value pruning and early filtering strategies are adopted to reduce the number of judgments in the ranking calculation process and further optimize the query efficiency.The correctness and effectiveness of the algorithm are verified on the simulated and real datasets.Experiments show that the query method proposed in this thesis can solve the problem of the result set quality and query efficiency simultaneously.(2)In order to improve the usability for Reverse k-Ranks Query results,and help users get the desired query result,we analyze the Why-not problem in Reverse k-Ranks Query and construct the cost model of the query adjustment.On this basis,a novel algorithm called MWKR is proposed to solve the Why-not problem in Reverse k-Ranks Query.The query adjustment strategy of modifying the missing vector set and k value is adopted,and finally,we can obtain the approximate optimal solution by means of sample space clipping and result construction.The validity and efficiency of the algorithm are verified on the simulated data set and the real data set.(3)Based on the real application scenarios of Reverse k-Ranks Query,a visualization system supporting Reverse k-Ranks Query under spatial constraints is designed and implemented.Firstly,we analyze the query requirements in real application scenarios,and propose a query model and data model.Then,the overall architecture,data exchange module,index module and query engine module of the system are introduced in detail.On this basis,the prototype system is implemented by Java Web technology,and the query process and result presentation are presented in combination with a real user query scenario,thus verifying the availability of the system.The user-friendly front-end visual interface allows the query results to be dynamically rendered on the map and parallel axes,which enables the users to analyze the results intuitively.