International Research Journal of Engineering and Technology (IRJET) Volume: 11 Issue: 11 | Nov 2024
www.irjet.net
e-ISSN: 2395-0056 p-ISSN: 2395-0072
DETECTION OF POLITICAL HATE SPEECH FOR SOCIAL MEDIA MONITORING Vaibhav Shah1 and Soham Kulkarni2 and Chinmay Deodhar3 and Viraj Shah4 and Priyanca Gonsalves5 1 Dwakadas J. Sanghvi College of Engineering, Mumbai, India 2 Dwakadas J. Sanghvi College of Engineering, Mumbai, India 3 Dwakadas J. Sanghvi College of Engineering, Mumbai, India 4 Dwakadas J. Sanghvi College of Engineering, Mumbai, India 5 Dwakadas J. Sanghvi College of Engineering, Mumbai, India
---------------------------------------------------------------------***--------------------------------------------------------------------undermine social unity, and silence those who are already marginalized. With the growth of online platforms and social media, the reach and influence of political hate speech have increased, making it more critical than ever to create effective strategies for detection and mitigation. The main goal of detecting political hate speech is to recognize and categorize instances of this harmful language across different forms of communication, such as social media posts, news articles, and public speeches [8].
Abstract - Political hate speech is an expression that is conveyed from a person or a group towards other person or group as an attack using various media. It can be found in a variety of online and offline contexts, including social media, news articles, and public speeches. Political hate speech can result in various negative consequences like public unrest , riots and protests and increased criminal activity. This project will develop a novel approach to political hate speech detection that combines machine learning and NLP techniques. The approach mentioned in this work would identify and note the features that determine the hate speech. These features may include keywords, phrases, and sentiment. After the features are successfully extracted from the raw text , they are input to the machine learning models which will predict if the speech contains hate. The input data can be in any format like text, audio files, or even YouTube URLs and it will classify what type of emotion and the percentage of hate speech comes from it. The machine learning model mentioned in the work will be trained on a large publicly labelled dataset. The developed model will also predict the emotion of the given input sample and classify it to various parameters like joy , sadness, anger ,surprise , etc. Our proposed work will be based on a text dataset of political speeches and articles. The results of the work will be reviewed in terms of accuracy and flawlessness of the model to predict the correct label. The proposed approach will also be compared to existing hate speech detection systems. In this work we have implemented the task in different machine learning algorithms including adaboost , decision tree and random forest [1] [3] [5].
Fig. 1. Active user on social media We can: Prevent and protect people or groups from harm, Prevent cyberbullying from occurring in the first place, help the police and courts: Hate speech can be traced in order to assist both the law enforcement and legal process in bringing a perpetrator of hate crime to justice [2]. In a nutshell, political hate speech reaches even the most sensible of individual's modes of detection, which in itself proves that individuals can be protected further. This has been interpreted to mean support for further investigation of the political hate speech through novel means that perpetuates further discussions of such continuations. [5].
Key Words: Political hate speech, Emotion Detection, Automatic detection, Machine learning, Natural language processing (NLP), Feature extraction, Sentiment analysis, Classification.
1. INTRODUCTION The widespread nature of political hate speech presents a serious danger to individuals, communities, and democratic societies. This damaging type of expression can provoke violence,
© 2024, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 68