CNN based Hate-o-Meter: A Hate Speech Detecting Tool
Ajinkya Chaudhari, Akshay Parseja, Akshit Patyal · 2020 Third International Conference on Smart Systems and Inventive Technology (ICSSIT) · 2020
Hate Speech is a widespread problem that degrades a person or people based on their race, religion, gender or disability. This research work proposes a tool to raise awareness on the persistent hate speech in blogs, online-forums, and newspapers. The primary aim of this research work is to highlight the content that promotes violence or hatred against individuals or groups based on religion, gender, ethnicity or disability. A convolutional neural network architecture is used along with the natural language processing techniques. Using this algorithm, the tool identifies the percentage of hate and displays the bias of the statements. To host the proposed model, flask API and heroku platform is used. The proposed tool has the ability to detect hate speech with 80.15 percent accuracy and f1-score of 80.35 percent. The tool is made free and available for demo use to the public.