Stigma and systemic distrust: a multidimensional sentiment analysis of social media discourse on elder abuse in Chinese long-term care institutions

This study analyzed Weibo and Douyin comments posted during a five-day period surrounding a suspected elder abuse incident in a long-term care institution in mainland China. Of 51,776 comments collected, 41,362 remained after preprocessing. We used lexicon-based emotion classification, Baidu AI Cloud sentiment scoring, and TF-IDF and Latent Dirichlet Allocation to examine emotional, temporal, regional, and thematic patterns. Sentiment was predominantly negative (mean = -35.941). Among 4,305 classified comments, resentment was most frequent (n= 1,433, 33.29%), followed by confidence (n= 1,297, 30.13%), vengefulness (n= 659, 15.31%), fear (n= 538, 12.50%), and shame (n= 378, 8.78%). Higher provincial resident-to-direct care worker ratios were associated with more negative sentiment, although this ecological correlation does not establish causality. Recurring themes concerned prison-like images of institutional care, direct care worker stigmatization, regulatory oversight, vulnerability among people aging without children, and tensions surrounding family responsibility.