Traditional algorithm for mining association rules need to scan the database many times when mining association rules, which are inefficient and time-wasting. In the light of the defects of traditional algorithm, this paper introduces the improvement of community partition algorithm into the process of association mining and uses the Internet forum users’ data of a university as the object of the new algorithm for mining association rules. The experiment shows that the new algorithm can help optimize the data association rules mining and reduces the times and time of scanning the database, so the mining efficiency is greatly enhanced.