International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 09 | Sep 2024
p-ISSN: 2395-0072
www.irjet.net
An AI System for the Detection of Hate Speech Encoded in Igbo Native Language P. Ana1, G. I. Emereonye2, A. C. Onuora3, C. C. Ukaegbu4, R. C. Aguwamba5, P. C. Sunday6 1Department of Computer Science, Cross River State University of Technology – Calabar, Cross River State, Nigeria. 4Department of Computer Science, Milwaukee school of Engineering University, USA.
5Department of Office Technology and Management, Akanu Ibiam Federal Polytechnic Unwana, Ebonyi State,
Nigeria.
2,3,6 Department of Computer Science, Akanu Ibiam Federal Polytechnic Unwana, Ebonyi State, Nigeria.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - This project was motivated by the pressing need
To combat this issue, social media giants like X and Meta have implemented several strategies. They have established comprehensive policies against hate speech, clearly defining its boundaries and outlining the consequences for violators [1]. Furthermore, they utilize advanced technologies such as machine learning algorithms and human reviewers to detect and remove hateful content. Additionally, social media platforms have implemented measures like user identity verification and reporting tools to enhance accountability and facilitate user involvement in combating hate speech.
for language-specific solutions to combat hate speech nuances in Igbo native language. The aim is to develop a robust Hate Speech Detection System integrated into the Facebook platform. The specific objectives include creating the first-ever hate speech dataset for Igbo, employing advanced natural language processing (NLP) techniques, and ensuring ethical considerations in system deployment. Methodologically, the project involves comprehensive data pre-processing, neural network model creation using Keras, systematic testing, and deployment on Facebook with a meticulous process. The key findings include the successful development of a reliable hate speech detection model with specific strengths and weaknesses. The project contributes to knowledge by establishing a systematic approach to language-specific hate speech detection and emphasizes the importance of ongoing efforts such as dataset expansion and community engagement.
Despite these efforts, hate speech remains a persistent challenge due to its evolving nature. Therefore, it is crucial to promote user education, ensuring awareness of hate speech and reporting mechanisms. Users and regulators must hold social media companies accountable for their role in combating hate speech. Supporting organizations dedicated to countering hate speech through donations or volunteer work can also make a significant impact. By integrating these multifaceted approaches, we can strive to foster a safer and more inclusive online environment, where hate speech is effectively addressed and minimized. This collective effort is essential to create a positive digital space for all users, promoting understanding, respect, and dialogue.
Key Words: Hate Speech, Igbo language, Artificial Intelligence, Transformer, Social Media
1. INTRODUCTION Hate speech is a denial of the values of tolerance, inclusion, diversity, and the very essence of human rights norms and principles. It may expose those targeted to discrimination, abuse, and violence, but also social and economic exclusion. When left unchecked, expressions of hatred can even harm society’s peace and development, as it lays the ground for conflict, tension, and human rights violations, including atrocity crimes.
Specific objectives of this study are: 1. Evaluating and validating the performance of a developed model using appropriate evaluation metrics and validation techniques. 2. Designing and implementing a software that can accept input of Igbo text and respond whether the text contains hate speech or not using the AI powered hate speech detection model.
Importantly, combating hate speech first requires monitoring and analysing it to fully understand its dynamics. Since the spread of hateful rhetoric can be an early warning of violence – including atrocity crimes – limiting hate speech could contribute to mitigating its impact. The authors of hate speech should also be held accountable, to end impunity. Monitoring and analysing hate speech is a priority for many UN entities, including UNESCO - the United Nations’ specialized agency for education, science and culture - which supports and undertakes research, which supports research to better understand its dynamics.
© 2024, IRJET
|
Impact Factor value: 8.315
3. Deploying and integrating the trained model into the application to enable detection of hate speech. The remaining sections of the paper are as follows: Section II provides a theoretical analysis of hate speech detection, covering definitions and relevant concepts. Section III outlines the methodologies used for software design and model training. In Section IV, the paper presents and
|
ISO 9001:2008 Certified Journal
|
Page 335