Research And Application Of Visualization Of Complicate Data

The improvement of communication technology and mobile network brings the increasing amount of communication data. Complex data of communication contains large amounts of information which reflect real social relationship and information flow. With these patterns and features, we can do a lot of work such as product recommendation and social security monitoring. But, there are too many communication network users and the data is becoming larger and larger which needs professional data process tool to analyze. In this thesis, we propose systematic research and engineering practice to solve above problems.1. Research visualization of multidimensional complex data. We visualize complex data to analyze it. The main visualization tool for complex data is undirected graph. In this graph, the participants of complex data are regarded as nodes and the relationships of data are regarded as edges(or links). At last, the whole data set is visualized into an undirected graph.2. Propose network intercommunication relationship extension algorithm based on network fusion and named entity discovery. A network’s intercommunication relationship represents the network’s connectivity. In a network with better intercommunication relationship, there are more paths between nodes and the analysis of network information flow and prediction work based on it are more accurate. Network intercommunication relationship extension algorithm based on network fusion fuses networks and adds mapping relationships to the network to extend intercommunication relationship. Network intercommunication relationship extension algorithm based on named entity discovery extracts named entity from network’s source data and adds them into original network to extend intercommunication relationship.3. Propose geographical based fast community discovery algorithm. A community in a network is in which nodes are connected dense. The traditional community discovery algorithm is strictly in accordance with the data relationship. Communities discovered in traditional way lost important offline information, the location information. Geographical based fast community discovery algorithm takes nodes’ location information into consideration to raise convergence rate and add the influence of nodes’ location to community distribution.4. Design and realize the complex data visualization system. This system firstly filters data that users are interested in and visualize it from different ways. Then analyze data using above algorithms. At last, calculate every network participant’s influence power. The users with high influence power are more suitable for conducting propaganda.