Analyzing Abusive Comments in Bangla: Machine Learning Study of Feminism on Social Media

The broad use of digital and Internet forums has recently facilitated various discussions and movements, including those centered on feminism. The escalation in online participation has concurrently given rise to a concerning surge in the use of offensive and disrespectful language targeting individuals advocating feminism. To address the prevalence of derogatory comments related to feminism on digital platforms, our study employed a machine learning process to delineate three specific categories of words in the Bangla language: positive, negative, and neutral terms. The primary objective is to establish an automated system proficient in accurately detecting and highlighting instances of abusive language, thereby cultivating a more constructive and harmonious online discourse. The proposed methodology employs natural language processing and machine learning techniques, commencing with the meticulous curation and annotation of an exhaustive dataset comprising 700 comments encompassing diverse forms of Bangla texts related to feminism, which is essential for the development of a robust and precise model that can yield meaningful insights. Ultimately, this study significantly contributes to improving automatic systems that can effectively and precisely identify abusive speech in the context of feminist comments.