International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 11 Issue: 07 | July 2024
www.irjet.net
p-ISSN: 2395-0072
Detecting Deception: A Comprehensive Review of Machine Learning Approaches to Fake News Detection Osama Zaheer1, Yash Bhagia 2, Khan Shah Ahmad Shakir Abu Asim3, Mohd Azhan Umar Kamil4 1,2,3,4Zakir Hussain College of Engineering and Technology, Aligarh Muslim University, Aligarh (202001) -U.P., INDIA
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - It was challenging to distinguish between false
the greatest threats to freedom of speech, the media, and democracy. It can even erode the public’s trust in governments. One notable instance of the effects of propagating fake news is the 2016 US Presidential Election. Gaining public’s trust and disseminating false information depend heavily on social and psychological aspects. Humans, for instance, have been found to become gullible and illogical in their ability to distinguish between truth and untruth when they are repeatedly exposed to erroneous information [1]. In this field of study, the application of machine learning algorithms, such as deep learning algorithms, graph-based techniques, multi-modal approaches, and explainable AI methods, have been concentrated on the detection of fake news. These developments in the sector greatly encourage future efforts to effectively tackle fake news. In general, fake news is detected by confirming videos and images as well as the identity of the sender by checking the account’s age on social media. Whether an account has blue account verification, validating the friends and followers as well as content that is frequently shared, mainly validates the URL. In order to solve this issue, machine learning algorithms have been developed for the automatic detection of fake news. These algorithms examine the trends and characteristics of both accurate and false news. On the basis of this information, they predict the validity of fresh publications. These algorithms can be trained using a large dataset to learn the traits and patterns of both real and fake news, recognizing the fake news with good precision and recall [2].
and true information due to social media networks’ easy access and rapid proliferation of content. The rapid spread of fake news has been facilitated by the simplicity of information dissemination on social media. Fake news has consequently grown to be a major issue with important repercussions for national security, public confidence, and media credibility. Thus, creating trustworthy and precise methods to identify fake information and stop its spread has become a big scientific challenge. To identify fake news, various machine-learning techniques have been proposed. However, the majority of those concentrated on a certain category of news (like political news), which raises the issue of dataset bias in the models employed. We want to provide a thorough analysis of the many methods for identifying false news using machine learning, which is essential in the classification of information. We also compared the performance of a variety of advanced pre-trained language models for fake news identification in addition to standard and deep learning models. Key Words: Natural Language Processing, Machine Learning, Deep Learning, Neural Networks, Convolution Neural Network
1.INTRODUCTION The widespread use of social media platforms has greatly impacted the dissemination of information, including news. The ability to access the most recent news has been made simpler for consumers by different online news platforms including social media, blogs, and other digital media formats. Approximately 68% of Americans as of August 2018 got their news from social media, up from 62% in 2016 and 49% in 2012 [1]. However, these platforms have also been used to spread fake news for monetary gain or to manipulate mindsets. Fake news has become a major issue in today’s society, significantly affecting how people think and make decisions, leading to erroneous beliefs, mistrust, and confusion. The main challenge in the existing content-based analysis is the difficulty in the detection of fake news, as it may encompass political, cinematic, and other issues. Therefore, common sense needs to be there for the detection of news datasets. The propagation of false information is a diverse topic on media platforms like Facebook, Instagram, and others. Fake news poses one of
© 2024, IRJET
|
Impact Factor value: 8.226
The complexity of the algorithms used for fake news detection varies; some rely on simple concepts, while others use complex deep learning models. One of the most popular techniques for examining news articles’ text is natural language processing (NLP). The text’s structure and substance, as well as the words used, their usage patterns, and their placement, can all be examined by NLP algorithms. Machine learning algorithms can also look at news sources, how they spread on social media, and other contextual information. Support vector machine (SVM), linear support vector machine (LSVM), K-nearest neighbor (KNN), logistic regression (LR), decision tree (DT), and Naive Bayse models combining SVM and LR are examples of machine learning (ML) models [3]. Another popular approach is to use network analysis to examine how news is shared on social media. This can
|
ISO 9001:2008 Certified Journal
|
Page 1228