Detecting Adolescent Psychological Pressures from Micro-Blog

Adolescents are experiencing different psychological pressures coming from study, communication, affection, and self-recognition. If these psychological pressures cannot properly be resolved, it will turn to mental problems, which might lead to serious consequences. Traditional face-to-face psychological diagnosis and treatment cannot meet the demand of relieving teenagers’ stress completely due to its lack of timeliness and diversity. With micro-blog becoming a popular media channel for teenagers’ information acquisition, interaction, self-expression, emotion release, we envision a micro-blog platform to sense psychological pressures through teenagers’ tweets, and assist teenagers to release their stress through micro-blog. We investigate a number of features that may reveal teenagers’ pressures from their tweets, and then test five classifiers (Naive Bayes, Support Vector Machines, Artificial Neural Network, Random Forest, and Gaussian Process Classifier) for pressure detection. We also present ways to aggregate single-tweet based detection results in time series to overview teenagers’ stress fluctuation over a period of time. Experimental results show that the Gaussian Process Classifier offers the highest detection accuracy due to its robustness in the presence of a large degree of uncertainty that may be encountered with previouslyunseen training data on tweets. Among the features, tweet’s emotional degree combining negative emotional words, emoticons, exclamation and question marks, plays a primary role in psychological pressure detection.