Employing TF-IDF and Neural Language Processing for Sentiment Classification of Potentially Radical and Hate Speech on Social Media

Employing TF-IDF and Neural Language Processing for Sentiment Classification of Potentially Radical and Hate Speech on Social Media

Akshay Patidar1, Dr. Sanmati Jain2

Research Scholar1, Associate Professor2

Vikrant University, Gwalior, India1,2

Abstract: Sentiment Analysis models are being used extensively as an important application of natural language processing. One such application is identification of hateful and radical speech on social media. With millions of posts and comments generated every day, relying solely on manual detection methods is impractical. Moreover, hate speech often evolves in subtle ways, making it difficult to track using predefined rules or simple keyword-based methods. This necessitates the development of automated systems capable of understanding and adapting to the nuanced and context-dependent nature of hate speech. The proposed approach combines TF-IDF and Neural Networks for identifying potential hate speech. The approach is based on the conjugate back propagation model, along with term and group frequencies as additional features. It has been shown that the proposed approach attains higher classification accuracy compared to existing work in the domain.

Keywords:- Natural Language Processing, Sentiment Analysis, Hate Speech, TF-IDF (Term Frequency–Inverse Document Frequency), Deep Learning, Classification Accuracy.