Reverse keyword-based location search on road networks

Reverse top-k keyword-based location query (RTkKL), aims to find the maximum spatial region such that the query object is contained in the result of any top-k spatial keyword query with users’ queried keywords and any location in the region as arguments. Existing efforts on RTkKL find the objects in the Euclidean space. In this paper, we study the problem of reverse top-k keyword-based location query on road networks. We propose two methods. One is based on mark vertex, and the other is based on bisector. For the mark vertex based method, we identify the mark vertex according to the definition of RTkKL on road networks. Based on the mark vertex, we will get the mark segments in the result. For the bisector-based method, we find the border points for the query q and some objects. With Dijkstra algorithm, we start from the query point q. For each closed edge, whose two adjacent vertices have been extracted from the min heap, we would search the border points on the edge, and count the border points for the adjacent vertex. For each method, we propose effective pruning strategy to reduce the search range and computation cost. Finally, experiments demonstrate the efficiency of the proposed algorithm.