Prediction model of interaction anxiousness based on Weibo data

Abstract

Adolescents who face social distress in real life are often accompanied by interaction anxiousness. To avoid direct social activities, they prefer to indulge in social networks to satisfy their psychological needs for interpersonal communication. Sina Weibo, China's leading social media platform, has a markedly young user base. It provides a rich sample of adolescents with interaction anxiousness and conditions for real-time monitoring. In this study, various word categories, such as perception of spatial distance and positional relationships, morality, and emotion, showed a significant relationship with interaction anxiousness. Furthermore, prediction models were established based on the original Weibo data of 839 active Sina Weibo users through a variety of machine learning algorithms to predict the scores of users' interaction anxiousness. The results showed that the performance of the prediction model established by the fully connected neural network was the best, and both criterion validity and split-half reliability were good (rcriterionvalidity = 0.30, rsplit − halfreliability = 0.76). This study confirms the validity of the prediction model of interaction anxiousness based on social media behavior data, provides a feasible solution to examine adolescents' interaction anxiousness, and provides a scientific basis for more targeted mental health interventions.