Data analysis algorithms for mining online communities from microblogs

摘要

Mining microblog data based on complex networks is conducive to the effective mining of useful information. This paper focuses on community mining. A complex network is introduced, followed by a community mining algorithm based on user similarity. Based on the similarity, different communities were divided, and experiments were carried out with real datasets. The experimental results showed that the accuracy of the algorithm was 87.5%, the recall rate was 87.1% and the operation time was 2.1 s. In the result of dataset 2, the average modularity of the designed algorithm was 0.532, which was better than the Girvan and Newman (GN) algorithm and there was no weak community structure, showing that the algorithm had better performance in community mining. The experimental results demonstrate the reliability of the mining algorithm and clarify the contributions of data mining for detecting communities from a microblog network.