Multiple-user closest keyword-set querying in road networks

Abstract

Location-based group queries have attracted increasing attention due to the prevalence of location-based services (LBS) and location-based social networks (LBSN). An important and practical application in these queries is the multiple-user closest keyword-set (MCKS) query that aims to search a set of Points of Interest (POIs) for multiple users in road networks. These POIs cover the query keyword-set, are close to the locations of multiple users, and are close to each other. This problem has been proved to be NP-hard. Unfortunately, existing solutions cannot handle this query efficiently and effectively. Specifically, the existing exact approach does not scale well with the network sizes and the existing approximation approaches, though scalable, have large error bounds. To address the above issues, a series of enhanced algorithms are proposed for the MCKS query problem in this paper. Specifically, a 3-approximation feasible result search algorithm is first proposed. Then, using the cost of the result returned by this algorithm as an upper bound, we present an efficient exact algorithm and an approximation algorithm with better performance guarantee. The exact algorithm is designed based on a set of efficient optimizations. The approximation algorithm improves the best-known approximation ratio from 157 to 1.5. Extensive performance studies with two real datasets demonstrate the effectiveness and efficiency of our proposed algorithms, which outperform existing algorithms significantly.