基于LDA模型和聚类算法的城市热点推荐与应用

LDA model and clustering algorithm in recommendation and application of city hot sites

针对新浪微博的短文本发表功能及地点签到功能,获取用户发布的信息。利用LDA(Latent Dirichlet Allocation)模型,将获取到的大量文本信息和地理位置进行分词和词频统计处理,从而获得签到的热点地理位置信息,并在地图上标注出来。在获得用户签到的位置信息基础上,合并约束搜索条件,利用多距离空间聚类算法,优化推荐功能,并向用户显示其周边诸如商场、景点、饭店等热门娱乐场所的具体地理位置信息,进行热点推荐。

According to the functions of short text posting and sign-in to elicit the details post by the users.Cutting the vast short texts and geography positions to the phrases by LDA（Latent Dirichlet Allocation） Model,in order to count up the frequency of every phrase,and then obtain the hot geography positions,as well as label them on the map.With the Spatial Distance Clustering Algorithm,optimizing the recommendation function when the users offer their situations and restrict the searching conditions.And the system shows the details of some active sites,such as shopping malls,hot sites and restaurants to recommend to the users.