Abstract: Objective To understand the characteristics and sources of negative emotions of wives of gay, to provide ideas and data support for mental health care and intervention, and to promote the mental health of wives of gays. Methods This study focused on the postsin “Bar of Wives of Gays”of Baidu Post Barsince October 14, 2021. Python was used to crawl the posts up to November 7, 2021, and the data of 72,680 posts were obtained. Through word segmentation and TF-IDF algorithm, the top 50 high-frequency keywords were obtained. Gephi software was used to analyze the co-occurrence network of keywords and the emotional dictionary provided by Boson to judge the positive and negative emotional degree of the extracted high-frequency keywords.The number of posts per hour was calculated, and the positive and negative emotional characteristics of the posts were analyzed. Results The morning 03:00 to 05:00 witnessed the lowest user activity (3.01%), and 20:00 to 23:00 in the evening the highest (23.04%). There were 29,414 posts with positive emotional expressions, including 29,386 posts with emotional score >0.5, accounting for 40.40% of the total, and 28 posts with emotional score=0.5 (0.03%). A total of 43,266 posts with emotional score<0.50 expressed negative emotions, accounting for 59.50% of the total. The average emotional score of each period was negative. When expressing negative emotions, “disgusting”, “doubt”, “anguish”, “unpleasure”and “damage”were the most mentioned words. The posts mainly focused on marriage. Conclusion Negative emotions from marital problems are frequently-seen in "Bar of Wives of Gays"of Baidu Post Bar. More attention should be paid to the mental health of wives of gays, and timely psychological nursing intervention is significant for this vulnerable group and the protection of the legitimate rights and interests of wives of gay should be strengthened.

Key words: bar of wives of gays, artificial intelligence, web crawlers, affective analysis, social media