Tweeting in Support of LGBT?: A Deep Learning Approach

Aparup Khatua, Erik Cambria, Kuntal Ghosh, Nabendu Chaki, Apalak Khatua

Abstract: This paper explores the Twitter deliberation as a response to the Supreme Court of India's verdict regarding the decriminalization of homosexuality. We have collected 0.58 million tweets to gauge the societal perception about the LGBT community whether social media users are supportive or skeptic about sexual minorities. Our deep learning algorithms (accuracy in the range of 85%) have outperformed machine learning algorithms (accuracy in the range of 60%). Furthermore, we note that dominant aspects of supporting tweets are revolving around equality, justice, human rights of the LGBT community. On the contrary, opposing tweets are perceiving this verdict as a threat to the Indian culture, tradition and the family system.