Affiliation:
1. University of Chinese Academy of Sciences, China
Abstract
As today's online social network (OSN) has become a part of our daily life, the huge amount of OSN behavior data could be a new data source to detect and understand individual differences, especially on mental aspects. Based on the findings revealing the relationships between personality and online behavior records, the authors tried to extract relevant features from both OSN usage behaviors and OSN textual posts, and trained models by machine learning methods to predict the OSN user's personality. The results showed fairly good predictive accuracy in Chinese OSN. The authors also reviewed the same kind of studies in more pervasive OSNs, focusing on what behavior data are used in predicting psychological profiles and how to use them effectively. It is foreseeable that more types of OSN data could be utilized in recognizing more psychological indices, and the predictive accuracy would be further improved. Meanwhile, the model-predicted psychological profiles are becoming an option of measurements in psychological studies, when the classical methods are not applicable.