Analyzing Trends and Topics of Sinhala Hate Speech on Twitter: A Time Series Approach

With the exponential increase in popularity, social media has become an influential platform for expressing opinions on various topics. However, these conversations may contain both hate-promoting and non-hate-promoting content, and excessive promotion of hate content can significantly impact the masses. Hence, it is critical to identify trending periods of hate-promoting topics to avoid harmful incidents among individuals or society. Considering the previous experiences in Sri Lanka, it would be helpful to detect such discussions early. This paper proposes a deep learning-based approach to identify trending periods of hate topics on Twitter, specifically focusing on hate-promoting Sinhala topics. The proposed model can identify trending periods of hate-promoting content in real-time, providing valuable insights for policymakers and stakeholders to counter the spread of hate speech. This research has significant implications for mitigating the impact of hate speech in the Sinhala-speaking community, contributing towards a more inclusive and tolerant society.