Direction Aware Collective Keyword Query On Spatial Database

With the popular of portable GPS devices, such as location enabled mobile phone,location-based geographic information service gradually been widely known and need.Demand leads to the development of research, spatial database is becoming a popular fieldof study, spatial keyword query is one of the research directions, and spatial keywordquery refers to a query taking into account the spatial position and the object keywords.Due to current research mostly concern about direction or collection, therefore, this paperproposes a query named direction aware collective spatial keyword query, we have twokinds of sets: one with central object; another without central object.Firstly, we proposed the direction aware collective spatial keyword query with centralobject. This query solve the problem of querying some sets satisfy some requirements in acertain direction. The central object of the set is an object contains specific keywords, theadditional objects surround the central object. During the query, we use the grid structureto realize the direction query, when a central object is found, we process the surroundingquery, aim to find all surrounding objects that are in the direction and cover all thesurrounding keywords, than we calculate the Cost of the set. This processing will berepeated until when k sets were found, than the processing will be ended. In thesurrounding query, we take advantage of the search boundary to realize the pruning ofsurrounding objects, so we can reduce the amount of computation, and accelerate theresearch speed.Secondly, we proposed the direction aware collective spatial keyword query withoutcentral object. The processing of this query is similar to the direction aware collectivespatial keyword query with central object, but the most important distinguish between thequerys is the processing of the surrounding query. In this query, we firstly verify twoborder objects of the set, than find all objects that could make up a set with the two borderobjects, after confirming that all the objects cover all keywords, we combinate all theobjects, now we have different combinations. This query is a NP problem, so we proposedan approximate algorithm and an accurate algorithm.Finally, we test the two search algorithms with random dataset to verify the effective and efficiency of the algorithms, also we prove it is so.