User behavior analyses based on network data stream scenario

Traditional data stream analysis focused mainly on the protocol and the content of the stream, which ignored the behavior information delegated by the stream scenario. This paper proposes a user behavior analysis method based on network data stream scenario. Aiming at the characteristic of network stream, we model the user´s network access behavior in formalization and express the user´s network access behaviors in data stream graph. On this basis, we abstract out user behaviors patterns by aggregating and contracting the data stream graph into clusters, identify the user behavior by matching the user´s access behavior and the user behavior pattern. The experiment results show that the method demand in user network behavior analysis is able to meet with lower false rate and missing rate.