AUTHOR=Yang Bing Xiang , Chen Pan , Li Xin Yi , Yang Fang , Huang Zhisheng , Fu Guanghui , Luo Dan , Wang Xiao Qin , Li Wentian , Wen Li , Zhu Junyong , Liu Qian TITLE=Characteristics of High Suicide Risk Messages From Users of a Social Network—Sina Weibo “Tree Hole” JOURNAL=Frontiers in Psychiatry VOLUME=13 YEAR=2022 URL=https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2022.789504 DOI=10.3389/fpsyt.2022.789504 ISSN=1664-0640 ABSTRACT=Background

People with suicidal ideation post suicide-related information on social media, and some may choose collective suicide. Sina Weibo is one of the most popular social media platforms in China, and “Zoufan” is one of the largest depression “Tree Holes.” To collect suicide warning information and prevent suicide behaviors, researchers conducted real-time network monitoring of messages in the “Zoufan” tree hole via artificial intelligence robots.

Objective

To explore characteristics of time, content and suicidal behaviors by analyzing high suicide risk comments in the “Zoufan” tree hole.

Methods

Knowledge graph technology was used to screen high suicide risk comments in the “Zoufan” tree hole. Users' level of activity was analyzed by calculating the number of messages per hour. Words in messages were segmented by a Jieba tool. Keywords and a keywords co-occurrence matrix were extracted using a TF-IDF algorithm. Gephi software was used to conduct keywords co-occurrence network analysis.

Results

Among 5,766 high suicide risk comments, 73.27% were level 7 (suicide method was determined but not the suicide date). Females and users from economically developed cities are more likely to express suicide ideation on social media. High suicide risk users were more active during nighttime, and they expressed strong negative emotions and willingness to end their life. Jumping off buildings, wrist slashing, burning charcoal, hanging and sleeping pills were the most frequently mentioned suicide methods. About 17.55% of comments included suicide invitations. Negative cognition and emotions are the most common suicide reason.

Conclusion

Users sending high risk suicide messages on social media expressed strong suicidal ideation. Females and users from economically developed cities were more likely to leave high suicide risk comments on social media. Nighttime was the most active period for users. Characteristics of high suicide risk messages help to improve the automatic suicide monitoring system. More advanced technologies are needed to perform critical analysis to obtain accurate characteristics of the users and messages on social media. It is necessary to improve the 24-h crisis warning and intervention system for social media and create a good online social environment.