The Design And Implemention Of Distribute System For User Behavour Analysis

Information plays an increasing important role in our lives. Massive data analysis is more and more popular in the area of advertising and scientific decision-making. It’s widespread concern on how to find the valuable information from the data.With the remarkable innovations on the network bandwidth and user scale in recent years, carriers’ capability of networking management was challenged. The technology of network traffic monitoring is an effective method for them to analyze network statues. Network traffic monitoring system helps them to get massive data of network users, which implies the important information of network traffic and user behavior characteristics.But due to the large scale and complexity of the monitoring data, the user behavior analysis still needs improved. Distributed storage and computing system have developed rapidly in recent years, and it has gradually become an effective method for analyzing massive data. Combined with the format of network monitoring data, characteristic of traffic and analysis requirements, this thesis presents a User Behavior Analysis System (UBAS) based on the idea of distribution. UBAS composed with the data storage sub-system and data analysis sub-systems. Data storage sub-system, with distribute method and index structure, not only improved the storage capability but also have an efficient data retrieval performance. Data analysis sub-system provided a friendly interface to meet most of analysis requirements. The sub-system defined a strategy to control the compute process as well. Finally we analyzed the GPRS users’behavior in a Chinese city with UBAS.