Research On Users' Geographic Information Interest In Web Map

In the past decade,Web Map has experienced rapid development and greatly satisfied the geographic information needs of users;it is playing an increasingly indispensable role in people's daily life.The ensuing massive access log data contains abundant information and value,providing a good data foundation for exploring users' geographic information interests.Studying the distribution characteristics of users' geographic information behind their access behavior is helpful to understand users' geographic information needs in Web Map,so as to further improve the quality of the Web Map services.At present,the mainstream researches on users' geographic information interests start from the single aspect of user access content or access behavior,like the time distribution characteristics of group users' access behavior,the temporal and spatial variation characteristics of users' access content,and the response capability of servers to access requirements,to improve users' experience quality in Web Map.There is a lack of integration of user access behavior and access content in the scale of access session,and there is not enough exploration in the distribution of users' geographic information interest.In view of these shortcomings,this paper fully analyzes the multiscale features of users' access to Web Map.By integrating users' access behavior(such as access duration)and access content,VSGIIM(Vector Space Geographic Information Interest Model)– a geographic information interest model based on VSM is proposed,which can explore the distribution of users' geographic information interest in the process of Web Map access at session granularity.First,this paper preprocesses massive user access log data and POI(Point of Interest)data.IP-based user identification and time interval-based session recognition are used to obtain sessions of Web Map access;and attributes like central coordinate,layer,area of interest,access duration are extracted from each session.POI categories and map layers are associated based on the multi-scale display feature of Web Map;then the POI and its category distribution in users' area of interest are obtained in the current layer.This paper selects 16 POI categories as access content interest features of a session,while the access duration and time of the users' operation were recorded as access behavior interest features.Second,this paper proposes a method for calculating the geographic information interest degree which integrates user access content and access duration.On the one hand,the TF-IDF(Term Frequency–Inverse Document Frequency)method is used to obtain the initial interest degree based on the quantity distribution of user's access to different POIs categories in a session;on the other hand,by introducing the user's access time to different areas of interest,the initial interest degree is updated,and finally the user's interest degree on different interest categories of the content category is obtained.The proposed calculation method for interest degree can accurately describe the distribution difference of the user's geographic information interest in different access sessions.Finally,this paper proposes a geographic information interest model called VSGIIM,which aims to explore the distribution characteristics of user geographic information interest at session granularity.The interest feature of VSGIIM lies in the integration of users' access content and access behavior at session granularity.Through K-means clustering of user access sessions,the diversity and difference characteristics of geographic information interest in user access sessions were analyzed.In the end,this paper continued to explore the temporal distribution characteristics of different access session categories,and analyzed the divergence degree of users' geographic information interests;with two typical users selected out,this paper detected the change characteristics of high-frequency users' geographic information interest,for the evolution of their geographic information interests.This provides some basis for future researches related to content prediction in Web Map access.This paper conducts a series of experiments and analysis with real data set of Tianditu user access logs,which verifies the accuracy of VSGIIM in the description of users' geographic information interest,and provides some guiding significance for deeper researches on the users' geographic information interest in Web Map.