NLP-based Emotions Analysis of the Marginalized Communities of Bangladesh

Human life relies heavily on emotions, as they influence our decision-making, social connections, and daily conduct. Current research has primarily focused on predicting emotions from social media data, neglecting to explore the underlying reasons behind emotions among marginalized populations in developing countries. In this study, we explored total of a 108 marginalized people’s interview conversational text data of Bangladesh. The study aimed to understand the joy, fear, anger, and sadness emotions based on computational machine learning model of IBM Tone Analyzer and figure out possible reasons behind their individual emotions. Our study reveals that fear and anger were relatively low across the community. The study also finds that male laborers have higher sadness than women, low-income participants have higher joy emotions, young laborers have high sadness, and participants who live in rural areas are happier than those who live in urban areas. The study will assist future technology designers create emotion-aware technologies.