User Behavior Analysis And Function Optimization Of Enterprise Information System Based On Bipartite Network

With the increasing maturity of cloud computing technology, enterprise information systems delivery and use will be transformed into a service model, the design of enterprise information system that originally to function as the center proposed more non-functional requirements, such as user experience, proactive demand, and so on. While user's information system usage behavior bears facts and rules of users and information systems. Therefore, research on the usage of information system for optimizing information systems and service transformation has theoretical and practical value.Based on the enterprise information system, we log data as original data, build the “User-Function” bipartite network, analysis user's behavior and furthur research the relationship between roles and system modules, finally research intrinsic relationship between the user's behavior of information system and the structure of the system function, according to the community discovery method and many more. Specifically, the main research contents are as follows:(1) We build the user and function bipartite network based on the enterprise actual use data. Based on bipartite network model and enterprise information system, we use the log, after data preprocessing, build a "User-Function" bipartite network, and measure the indicators of the network, and the results indicate that the user is fully functional network connectivity, but there is a smaller network density, network overall standard of complex networks, and the exchange of information network is good.(2) We study the role behavior of information system. Based on user and functional networks, we gather the two types of nodes and get a new “Roles-Module” bipartite network. We analysis the difference of usage behavior across different roles, and analysis user of different roles found that users belong to the same role, but there are great differences between individuals, this difference is mainly in use modules and features of the type and quantity, frequency of use, the use of long, etc., which will help identify the key users of different roles.(3) We study the intrinsic relationship between the user's behavior of information system and the structure of the system function. Through the projection to the user and function network, according to the number of common users as the weight to get the functional relationship network, build communities of the mining model based on community level clustering algorithm and the method of module function and so on, we divided the community structure by this model, and analyzed and simulated the results of division, and verified the validity of the results. The results show that based on the functional structure of use division more fully reflects the consideration of the user experience.