Hate Speech Detection Network Using LSTM
Chirag Lala, Pulkit Dwivedi · 2023
Socia1 media is an extremely popular form of communication today. People offer their opinions and insights on a variety of topics, including politics, video games, and their personal lives. These platforms are occasionally used by some people to propagate false information about another person or group of people. The term “hate statement” refers to this kind of offensive material. One of the most well-known social media platforms is Twitter. But many people also use Twitter to disseminate offensive material. It is very hard to manually weed out abusive comments from the hundreds of millions of tweets that are generated every day on Twitter. Therefore, these offensive tweets ought to be automatically filtered out. In this study, we are developing an LSTM model for categorising tweets as either containing hate content or not. The dataset used is a publicly available dataset on Kaggle. This model has an accuracy of 98.04% on the training dataset and 97.19% on the validation dataset. For the test dataset, a precision of 0.98 is obtained for non-hate tweets and 0.93 for hate statement tweets.