Identifying and Analyzing Cancer-Related Stigma on Chinese Social Media: A Corpus Analysis

Cancer stigma is a profound psychosocial burden (Lebel & Devins, 2008; Vrinten et al., 2019), yet its culturally-specific connotations and expression within Chinese online communities remain poorly understood. This study addresses this gap by proposing a novel methodological framework that triangulates rich interview data with large-scale naturalistic social media data to analyze stigmatizing language. First, we constructed a list of stigma lexicon drawn from indepth interviews of patients and caregivers (n=30). The lexicon was then used to screen a corpus of cancer-related posts on the platform Rednote. Each post was scored for its relevance to the stigma lexicon using TF-IDF weighted cosine similarity. A validated cutoff for stigmatizing content was established from a stratified sample of 100 posts manually coded by volunteers with cancer experience. Then, we conducted a comparative keyword analysis of identified stigma corpus against a non-stigma reference corpus. The analysis produced data-driven conceptual model explaining online expression of stigma. This research provides the first validated, scalable method for measuring culturally-contextualized cancer stigma in naturalistic Chinese social media settings. The findings offer a roadmap for public health surveillance and the development of targeted, culturally-informed digital interventions to mitigate the isolation and distress caused by cancer stigma.