MODIFIED TF-IDF WITH MACHINE LEARNING CLASSIFIER FOR HATE SPEECH DETECTION ON TWITTER

Abstract

Whether it's written, spoken, or symbolic, any kind of communication that targets individuals or
groups based on characteristics like race, religion, ethnicity, gender, sexual orientation, or disability is
considered hate speech. Twitter, with its massive user base and ease of communication, has become a breeding
ground for hate speech. Tweets are generated at an incredible rate, making it impossible to manually evaluate
and categorize them for hate speech. In order to identify potentially hateful content, many of the conventional
ways for doing so rely on lexicon-based approaches, where predetermined lists of offensive or discriminatory
terms are utilized. However, these approaches generally lack the context necessary to effectively discriminate
between hate speech and other types of expression, and they struggle to adapt to the ever-changing nature of
hate speech. Due to the inefficiencies of the currently available methods, cutting-edge strategies are required for
the automatic detection of hate speech on Twitter. Through the use of algorithms, machine learning classifiers
offer a viable answer by learning patterns and features from massive datasets. By using the TF-IDF method, we
are able to identify the specific features of hate speech and create a reliable model for identifying it.