利用手机定位数据的用户特征挖掘
Personal Profile Mining Based on Mobile Phone Location Data
-
摘要: 目的 手机定位数据已经逐渐成为一类新兴的空间数据,可用于分析个体或大规模区域内群体的活动特征,服务于基于位置的服务和城市及交通规划等。提出了一种基于手机定位数据,结合区域内兴趣点(POI)、房产价格等,利用空间聚类及语义分析等手段,对用户特征进行分析和挖掘的方法。首先采用DBSCAN方法提取用户重点活动区域;其次,根据用户的活动规律假设对活动区域进行类别标注;最后引入自然语言处理方法对POI和楼盘描述信息进行词频分析。并结合区域内POI类别和房价信息推断用户可能的偏好特征及收入或消费能力等特征,对用户一个月的手机定位数据进行挖掘分析。结果表明,该方法对用户重点活动区域及个体喜好特征等能够进行较为有效的挖掘。Abstract: Objective Understanding personal profiles like preferences,income levels,and geographical areas isthe basis of providing aperson with personalized and accurate services.In order to acquire personalprofiles we propose a reasonable technical route that first extracts the geographic regions from person-al mobile phone location data based on a density-based clustering algorithm.Then,the geographic re-gions are tagged with semantic meaning and we analyze house descriptions by NLP(Natural LanguageProcessing).A division method for people’s daily time is given,based on the assumed the activitypatterns of people.At last,an individual’s taste for something or his income levels is analyzed usingthe statistics for POIs and house prices in the extracted places.An experiment with real data showsthat this method is an effective solution to mining personal profiles.
下载:
