Density Based Collective Spatial Keyword Query

Geographic objects with descriptive text are gaining in prevalence in many web services such as Google map. Spatial keyword query which combines both the location information and textual description stands out in recent years. Existing works mainly focus on finding top-k Nearest Neighbours where each node has to match the whole querying keywords. A collective query has been proposed to retrieve a group of objects nearest to the query object such that the group's keywords cover query's keywords and has the shortest inner-object distances. But the previous method does not consider the density of data objects in the spatial space. In practice, a group of dense data objects around a query point will be more interesting than those sparse data objects. Inner distance of data objects of a group cannot reflect the density of the group. To overcome this shortage, we proposed an approximate algorithm to process the collective spatial keyword query based on density and inner distance. The empirical study shows that our algorithm can effectively retrieve the data objects in dense areas.