Keyword-aware optimal route queries are a combinatorial optimization problem with three factors, namely keyword coverage, route budget constraint and route popularity, which is NP-hard. Previous work takes an adjacent edge expansion approach to get an approximate result and the computational complexity is proportional to the number of vertices and edges of the road network, which is not scalable for large road networks or road networks with sparse points of interest. Motivated by this, we propose an algorithm called Keyword-aware Skyline Route Generating (KSRG*). KSRG* consists of a preprocessing stage, in which KSRG* computes all the skyline routes for each pair of vertices, and an expansion stage, in which KSRG* expands feasible routes from the source vertex to target vertex via vertices containing keywords. Based on KSRG*, another two approximation algorithms with performance guarantees, namely KSRG\(^{+}\) and KSRG\(^{++}\) are proposed. KSRG\(^{+}\) applies three acceleration strategies to KSRG* to greatly reduce the routes to be expanded. KSRG\(^{++}\) combines KSRG\(^{+}\) with a clustering strategy to solve the inefficient performance when the number of the vertices containing the same query keyword is large. Comprehensive experiments on the datasets of four real road networks show that the execution time and accuracy of results of KSRG\(^+\) and KSRG\(^{++}\) outperform the state of the art methods.

关键词感知最优路径查询是一个组合优化问题，具有三个因素，即关键词覆盖率、路径预算约束和路径流行度，是 NP-hard。以前的工作采用相邻边扩展的方法来得到一个近似的结果，计算复杂度与路网的顶点和边的数量成正比，这对于大型路网或兴趣点稀疏的路网是不可扩展的。受此启发，我们提出了一种称为关键字感知天际线路线生成 (KSRG*) 的算法。KSRG* 包括一个预处理阶段，其中 KSRG* 计算每对顶点的所有天际线路线，以及一个扩展阶段，其中 KSRG* 通过包含关键字的顶点扩展从源顶点到目标顶点的可行路线。基于 KSRG*，\(^{+}\)和 KSRG \(^{++}\)被提出。KSRG \(^{+}\)对 KSRG* 应用了三种加速策略，大大减少了要扩展的路线。KSRG \(^{++}\)将 KSRG \(^{+}\)与聚类策略相结合，以解决包含相同查询关键字的顶点数量较多时性能低下的问题。对四个真实道路网络数据集的综合实验表明，KSRG \(^+\)和 KSRG \(^{++}\)的执行时间和结果的准确性优于最先进的方法。