Association of the rule mining is a significant technique for characterizing network traffic behavior. However, there are still three obstacles to mining association rules as of network traffic data as efficiency with huge quantity of the results, and the insufficiency in direction of represent network traffic behavior. To engage in these problems our paper proposes a hierarchical rule mining with an approach for the granules mining associations. The proposed approach utilizes a top-down rules mining methodology, employing a subjectively specified rules template for hierarchies to generate interesting rules. This approach significantly enhances the efficiency of rule creation by allowing users to filter out uninterested rules based on their subjective criteria. The method also suggests pruning an original category of redundant rules that this work describes to reduce the quantity of laws. Ultimately, the approaches incorporate the idea of variety, which seeks to select interesting rules to enhance understand network traffic behavior. The experiment conducted
on traces of MAWI the network demonstrates the quality and efficiency of the framework suggested.