Abstract:

Social networks provide people a platform for expressing their own emotion,viewpoints,interest and suggestion.The users’texts,behaviors and social circles bring new challenges for data mining.In this article,we propose a new method through which we can classify and predict micro-blog users’MBTI personality values by utilizing their shared microblog information. Firstly, based on the analysis of microblog users’ information, texts and nontext features,which can indicate users’psycholinguistic and behavioral characteristics,are extracted.And then three classification methods—AdaBoost decision tree,support vector machine and Bayesian logistic regression,are adopted to classify the personalities of micro-blog users.Extensive experimental results indicate that the prediction accuracy are between 75% and 90% for the four different personality dimensions of MBTI.

Key words: social network;micro-blog;personality classification;AdaBoost decision tree;support vector machine;Bayesian logistic regression