An approach to automation of user's profile analysis
Information from users’ profiles on social networking sites is an important data source for analysis of the users’ psychological characteristics. Texts, video and audio files, images, public pages can be easily accessible and analyzed. We consider the ways of estimating the users’ psychological characteristics on the base of his or her profile in the social network VKontakte. We compare different machine learning models for the analysis of user's texts, such as linear regression, decision trees, random forest, support vector machine with linear, radial and sigmoidal kernels. Also we discussed the possible further stages of research including the sentiment analysis for better text description, the analysis of profile photo, and, finally, ways of combining all steps for estimating psychological characteristics of social networks users.