Detection of IAD Based on Personality Questionnaires of Chinese College Students and SVMs

The unprecedented development of the Internet brings many benefits but it also manifests the Internet addiction disorder (IAD) which has a huge impact on some people in our society. However, it is hard to diagnose and prevent IAD. In this study, we propose a new way to detect IAD using C-SVM and nu -SVM by inviting 580 Chinese College Students to complete the questionnaires which consist of Brief Self Control Scale (BSCS), the 11th version of Barratt Impulsiveness Scale (BIS-11), Chinese Big Five (CBF) and Chen Internet Addiction Scale (CIAS). BSCS, BIS-11 and CBF evaluate one's personality in nine subscales. CIAS score is used to label the high and low group. We compare the performances of two SVMs on two datasets with different features by 5-fold cross validation after normalization. It is found that IAD could be detected from personality questionnaire data using SVM effectively. In particular, C-SVM with RBF function on the dataset with only subscales normalized in the range of [-1,1] was preferred.