Abstract: 【Purpose/significance】To study the sentiments of onlookers towards different types of privacy leakage events in social me?
dia platforms, so as to determine the attitude of onlookers on different types of privacy disclosure, so as to promote social media plat?
forms to provide users with better services under the condition of user privacy security.【Method/process】Based on the four types of
privacy leakage incidents, a sentiment analysis framework combining Baidu sentiment tendency analysis and LDA topic recognition
was constructed. On four kinds of social media platforms , the comments of onlookers in typical cases of four types of privacy leaks
were captured and processed. The topic extraction is conducted by LDA topic recognition model, and sentiment classification is done
through the analysis of Baidu sentiment tendency analysis. And the emotional contrast analysis of the four types of privacy leak is con? ducted.【Result/conclusions】Generally speaking, about 75% of onlookers have negative sentiments to privacy leaks, of which the negative sentiments of onlookers about social privacy disclosure account for up to 82%. Specific to the onlookers on each platform, the onlookers on WeChat Official Accounts and Weibo have more negative sentiments, the onlookers on Zhihu have more neutral and nega? tive sentiments, and the onlookers on the TikTok platform account for a higher percentage of positive sentiments. According to the re? search results, some suggestions are put forward【. Result/conclusion】This paper creatively discusses the emotional differences of on? lookers on four kinds of social media on four kinds of privacy disclosure events, and the research on the emotional situation and inten? sity dimension is not deep enough.