Diagnosis of social network mental disorders (SNMD) is problematic since self-reporting questionnaires (massively applied in psychology) give subjective insight into the problem, which affects the accuracy of the measurement. Methods of measuring phenomena such as social network mental disorders must be reconsidered.
The authors introduce a new and effective method of identifying SNMD by mining online social data. They exploit features extracted from social network data as the indicators of the diagnostic criteria of SNMD. Users’ real behaviors on the Internet are used to discriminate between three types of SNMD: cyber relationship addiction (CR), information overload (IO), and net compulsion (NC).
The authors also proposed an SNMD-based tensor model (STM) to improve the performance of detecting SNMD. Particularly noteworthy indicators of SNMD are identified by the authors, such as parasocial relationship (PR), online and offline interaction ratio (ONOFF), social capital (SC), social searching versus browsing (SSB), temporal behavior features (TEMP), usage time (UT), disinhibition-based features (DIS), profile features (PROF), and feature effectiveness. They are noteworthy because they are indicators of SNMD, which can be observed on the Internet.
The authors evaluated the effectiveness of their proposed solution on a sample of 3,126 subjects. All the indicators reached the effectiveness of 83.1 percent, and STM appeared to be the best classification technique with accuracy of 89.7 percent. This method had the best accuracy among other classification techniques such as CF, Tucker, DTCVM, and TSVM.
This paper is recommended reading for everyone who does research on Internet addiction and Internet disorders.