A Detection Model For Identification Of Depressed College Students On Weibo Social Network

Depression has become one of the most widespread psychological problemsmankind facing at present. In2012, The World Health Organization(WHO)estimated the quantity of depressed people all over the world to be0.3billion.College students are vulnerable to attack, for lack of life experience and endurancecapacity meanwhile bearing too much responsibility. The purpose of this paper is tofind a way to recognize depressed users by mining the online data of weibo users.The main contents of this paper are:Firstly, we made a contrastive analysis of characteristics of depressed users andnormal ones.142depressed user and274normal ones were marked artificially,whose basic information and tweets were crawled. This paper then analyzed thedifference between depressed and others in the way they speak and act, and affirmedthat the frequency of words and emoticons were important for classificationrecognition.Secondly, we established a detection model for identification of depressed users onweibo social network. Based on frequency of words and emoticons in the tweets, thepaper classified users as depressed and normal ones with Artificial Neural Network,Deep Belief Network and Support Vector Machine. Of these three models, the bestwas adopted as the classifier, and its availability was validated by classifying testingsamples.Finally, we got1502weibo users who were college students artificially, and crawledtheir data. With the data, this paper investigated the general situation of depressionin college students, and based on the detection model it proposed. Then the situationwas described and analyzed. And some suggestions for university management werethrown out.