Text mining for identifying the nature of online questions about non-suicidal self-injury

Abstract

Abstract Objective The internet provides convenient access to information about non-suicidal self-injury (NSSI) owing to its accessibility and anonymity. This study aimed to explore the distribution of topics regarding NSSI posted on the internet and yearly trends in the derived topics using text mining. Methods We searched for the keyword “non-suicidal self-injury” (Ja-Hae in Korean) in the Naver Q&A using the statistical package R. We analyzed 7893 NSSI-related questions posted between 2009 and 2018. Text mining was performed using latent Dirichlet allocation (LDA) on the dataset to determine associations between phrases and thus identify common themes in posts about NSSI. Results In the LDA, we selected the following 10 most common topics: anger, family troubles, collecting information on NSSI, stress, concerns regarding NSSI scarring, ways to help a non-suicidal self-injurious friend, depression, medical advice, ways to perform or stop NSSI, and prejudices and thoughts regarding non-suicidal self-injurious people. Conclusions This study provides valuable information on the nature of NSSI questions posted online. In future research, developing websites that provide NSSI information and support or guidance on effectively communicating with NSSI is necessary.

Keywords

• Non-suicidal self injury
• Internet
• Data mining