With the development of mobile positioning and the popularity of potable mobile devices,Location-Based Services have becoming more and more popular,because spatial keyword query can meet users' demands better comparing with traditional queries.For example,when a user expresses his or her query preferences(e.g.,keywords),the system is supposed to return best kspatio-textual objects ranked according to the proximity to the query location and the relevance to the query keywords.To our knowledge,almost all the existing proposals for answering spatial keyword queries are only focused on single query point scenario.However,in some cases,there may be multiple queries.Therefore,we propose multi-source spatial keyword Top-kquery,and discuss this kind of query in Euclidean space and road networks.In Euclidean space,we propose threshold-based algorithm(TA),which first performs incremental Top-kspatial keyword query for each query point and then combines their results.Next,we propose another more efficient algorithm by treating the whole query set as a query unit(MBRA),which can significantly reduce the objects to be examined,and thus achieve higher performance.Besides,this algorithm uses best-first(BF)method to traverse the indexing structure.In road networks,we first introduce the structure of networks,which can significantly increase the efficiency of accessing the road networks,and then based on one query point situation,we describe the search algorithm called for single query point scenario.Finally,we propose for multisource scenario.Extensive experiments using real datasets demonstrate that our approaches are efficient in terms of runtime and I/O cost. |