Key Technologies Of Social Computing With User Behavior Dynamics In Mobile Networks

The mobile Internet era is developing rapidly,and data is an important link connecting society and network.Mobile phone online records not only contain a large number of users' temporal and spatial information,but also record their life status in the Internet in detail.Therefore,users' online behavior has become an important entry for fine-grained understanding of human behavior dynamics.At the same time,with the accelerating urbanization process and the increasing pressure of urban supervision,the research and supervision based on the dynamic behavior of Internet access has become the research focus of social computing applications in cities.Research and supervision based on mobile Internet user behavior can capture user information in time,space and content dimensions.Specifically,the dynamic monitoring of user access content and user space activities is the key technology of social computing based on mobile Internet user behavior.Starting from the background of mobile big data,this paper finely excavates the pattern characteristics and potential connections of user behavior in time,space and content dimensions from the perspective of urban supervision.The main work is as follows:1.The Usage Detail Records(UDR)of mobile Internet users is analyzed.The diversity and predictability of content dimension and location dimension in mobile Internet users' access behavior are studied,and the granularity of user behavior research is also explored.2.A dynamic research method of user access content based on tensor decomposition is proposed.The method considers that user's Internet behavior is the weighted sum of many typical basic characteristics.The context transfer tensor is constructed by using user's access content time series,and the reliability is extracted by using improved high order Singular Value Decomposition(AHOSVD).The context behavior of mobile users constitutes a feature basis,which represents the typical access content pattern of a specific group of people.In addition,combined with spatial and geographical factors,reliable spatial structures of online behavior of mobile users in both places are extracted.Because of the consideration of higher dimension user characteristics,the research results can provide a comprehensive description of user's online sequential behavior patterns,and can be more effectively used for resource allocation and supervision of urban managers.3.A dynamic user space discovery method based on graph signal model is proposed.Base stations are used as spatial nodes,and spatial dependencies among nodes are modeled as spatial network diagrams.Combined with the status of base stations at each time point,the spatial network diagrams are superimposed metaphorically to combine the spatial and temporal connections of base stations.The signals on the graph of node traffic are loaded and the graph signal model is constructed to reflect the spatial flow trajectory of users.The spectral wavelet operator is applied to the space.Wavelet coefficients are generated on different wavelet scales in graph signals.Through the analysis of wavelet coefficients,we can infer valuable information about the origin,spread and span of population flow.The above methods provide a new perspective for the study of human behavior dynamics from the three basic dimensions of time,space and content.The research results of key technologies of social computing based on mobile Internet user behavior provide effective guidance for urban infrastructure construction and supervision.