International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 12 | Dec 2024
p-ISSN: 2395-0072
www.irjet.net
Advancing News Veracity Prediction with NLP and Multimodal Approaches 1Rambarki Sai Akshit, 2Konduru Hema Pushpika, 3Rambarki Sai Aashik, 4Dr. Manda Rama
Narsinga Rao, 5J. Uday Shankar Rao, 6S. Manjith 1,5,6U.G. Student, Department of Computer Science Engineering, GITAM University, Visakhapatnam, India 2M.S. Student, Department of Data Science, University of Maryland Baltimore County, USA 3M. Tech Graduate, VLSI Design, GITAM University, Visakhapatnam, India 4Professor, Department of Computer Science Engineering, GITAM University, Visakhapatnam, India
------------------------------------------------------------------------***-----------------------------------------------------------------------Natural Language Processing (NLP), a subset of artificial Abstract - This research leverages the Fakeddit dataset,
intelligence (AI), offers a promising solution to this challenge. NLP enables machines to process, analyze, and understand human language, providing the tools to detect subtle linguistic patterns and signals that distinguish authentic content from fabricated material. Researchers in this field are leveraging advanced techniques to build robust systems capable of addressing the complexity of misinformation.
a comprehensive multimodal collection of over 1 million news samples with diverse labels for binary and finegrained classifications, to address the pressing challenge of fake news detection on social media. By integrating advanced Natural Language Processing (NLP) models with multimodal data, including textual content, user comments, and images, this study achieves superior performance in detecting misinformation. Using the transformer-based BERT model, the approach effectively captures subtle contextual nuances, attaining 93.8% accuracy. When combined with Support Vector Machines (SVM) in an ensemble, classification accuracy improves to 95.8%. Further incorporating multimodal features, such as image captions and metadata, enhances the model’s performance, with a text-image system achieving a peak accuracy of 96.8%. This work highlights the potential of multimodal integration and NLP techniques in developing robust systems to combat disinformation and maintain the integrity of digital discourse.
This study undertakes a comprehensive exploration of NLP-based approaches to detect fake news, with a focus on the latest advancements in machine learning and deep learning. By integrating large and diverse datasets, this research aims to not only identify instances of fake news but also unravel the linguistic and contextual factors behind their creation and dissemination. A multifaceted methodology is employed, encompassing various datasets from social media platforms to train and evaluate a range of NLP models. These models are designed to extract and analyze features such as sentiment, stylistic patterns, and linguistic structures.
Key Words: Fake News Detection, Natural Language Processing, BERT, LSTM, Multimodal Data Integration
Ensemble
Methods,
Beyond text-based analysis, this research also incorporates multimodal data, including images, captions, and user metadata, to enhance detection accuracy and reliability. By combining these diverse sources of information, the study adopts a holistic approach to the problem of fake news detection. Additionally, ethical considerations such as fairness, transparency, and potential biases are integrated into the research process, ensuring that the detection methods align with democratic values and do not compromise freedom of speech.
1. INTRODUCTION The rise of social media platforms has revolutionized how information is shared and consumed, granting unprecedented global reach to individuals and organizations alike. While this democratization of information has empowered people to voice diverse perspectives, it has also exacerbated a critical issue: the proliferation of fake news. Fake news, defined as deliberately false or misleading information, poses significant risks to public opinion, political stability, and even public health. As a result, developing effective methods to identify and curb the spread of misinformation on social media has become a pressing concern.
© 2024, IRJET
|
Impact Factor value: 8.315
In essence, this investigation aims to contribute to the growing body of knowledge on combating misinformation in the digital age. By utilizing cuttingedge NLP methodologies, diverse datasets, and ethical principles, this study seeks to strengthen the integrity of digital communication and mitigate the harmful effects of disinformation. The subsequent sections will delve
|
ISO 9001:2008 Certified Journal
|
Page 860