基于标签传递的地图区域化搜索

Map region search based on label propagation

地图搜索是搜索应用的一个重要分支。针对如何提供更加多样性的服务(例如区域化搜索),提出了一种基于标签传递的地图区域化搜索算法,通过将分散的兴趣点构建成邻接拓扑图,并以此传递关键字权重,进而从局部权重极值点拓展形成关键字密集子图,从而实现地图区域化搜索。同时根据用户位置,考虑距离因素将密集子图重新排序,从而实现个性化推荐。最后,实现了一个搜索引擎的原型系统,并在真实数据集上验证了算法的有效性。

Map search is an important area of the search applications.Aiming at how to provide more variety of service,such as regional search,a map region search algorithm based on label propagation has been proposed.The scattered point of interests(POIs)are constructed as the adjacency topological graph in order to propagate the weight of keywords.Then,the dense subgraphs of keyword is formed from local optimal points to realize map region search.At the same time,according to the distance between users' positions,the sub-graphs are reordered to achieve personalized map region search.Finally,a search engine prototype system is built to verify the validity of this algorithm on real dataset.