Abstract: In recent years,as a hotspot in data mining of physical network,social network data mining has made grati-fying achievements in the current user behavior analysis,interest recognition and product recommendation.With the advent of social networking business opportunities,many malicious users and malicious behaviors have also emerged,which have a great impact on the effectiveness of data mining.A malicious user identification method based on user behavior feature analysis was proposed.This method uses the principal component analysis(PCA) to mine the user behavior data in microblogging network,and ranks the weight of each feature.It can effectively identify malicious users with first six-dimensional principal component features.The new features fitted by the principal component features are used to improve the recognition performance of the system.The experimental results show that the proposed method can effectively sort the microblogging user features and identify the malicious users in the microblogging social network,which provides a good data cleaning technique for social network data mining in other directions.

Key words: Feature extraction, Machine learning, Malicious users, Microblogging, Principal component analysis(PCA)