PROJECT TITLE :

Detecting Stress Based on Social Interactions in Social Networks - 2017

ABSTRACT:

Psychological stress is threatening folks's health. It is non-trivial to detect stress timely for proactive care. With the popularity of social media, individuals are used to sharing their daily activities and interacting with friends on social media platforms, making it feasible to leverage on-line social network knowledge for stress detection. During this paper, we notice that users stress state is closely connected to that of his/her friends in social media, and we use a giant-scale dataset from real-world social platforms to systematically study the correlation of users' stress states and social interactions. We have a tendency to first define a group of stress-connected textual, visual, and social attributes from numerous aspects, and then propose a unique hybrid model - a issue graph model combined with Convolutional Neural Network to leverage tweet content and social interaction data for stress detection. Experimental results show that the proposed model will improve the detection performance by half-dozen-9 p.c in F1-score. By more analyzing the social interaction knowledge, we tend to also discover many intriguing phenomena, i.e., the quantity of social structures of sparse connections (i.e., with no delta connections) of stressed users is around fourteen percent more than that of non-stressed users, indicating that the social structure of stressed users' friends are less connected and simpler than that of non-stressed users.