Bangla Sectarian Expression Dataset

本研究开发了首个全面的框架，用于自动检测在线孟加拉语内容中的社区暴力标记，并伴随有最大的社交媒 体互动数据集（13K原始句子），这些互动属于四大暴力类别及其16种粗略表达。该数据集由社会科学家、语言学家和心理学家参与的7步专家注释过程构建，旨在监测和标记在线空间中的孟加拉语宗派文本，以应对南亚地区日益增长的社区暴力问题。This study develops the first comprehensive framework for automatically detecting community violence cues in online Bengali content, alongside the largest social media interaction dataset (13,000 original sentences). This dataset encompasses four categories of violence and their 16 coarse-grained expressions, and was constructed via a 7-step expert annotation process involving social scientists, linguists, and psychologists. It is designed to monitor and label sectarian Bengali texts in online spaces, aiming to address the escalating problem of community violence in South Asia.