International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 09 Issue: 07 | July 2022
p-ISSN: 2395-0072
www.irjet.net
A review of Fake News Detection Methods Sandhya S1, Arya J Nair2, Sreeja Kumari S3 1,2,3 Lecturer,
Dept. of Computer Engineering, NSS Polytechnic College, Kerala, India ---------------------------------------------------------------------***--------------------------------------------------------------------3. Chapter 4 depicts the comparison of these methods. Abstract – False news is the dissemination of false information to grab the audience's attention, which, at first glance, may seem credible. False news is frequently produced by people with personal, political, or economic objectives in mind. It is frequently disseminated online or through conventional media. Fake news can reach dozens and even millions of people via social media, even if it's labeled as fake. In order to lessen the impact of spreading misinformation as well as to eliminate the sources of this false information, all efforts should be made to reduce its impact on citizens. This paper examines and evaluates the many methods for identifying false news.
Chapter 5 includes the conclusion of the study.
Key Words: Convolutional Neural Network (CNN), Hybrid
A content-based fake news classification system has been proposed in [5] using the voting ensemble classifier method. Logistic Regression, Regression tree, ensemble AdaBoost, Naïve Bayes, Fuzzy Rule, KNN, Decision tree, and PNN are the voting classifiers utilized in this system. The model's classification accuracy for false news recognition was as high as 95%.
2. LITERATURE REVIEW A unique diffusive network-based false news authenticity checker has been introduced in [4]. A hybrid feature learning unit (HFLU) has been used in this system, which is trained to learn both obvious and implicit feature representations of news. The research suggests a unique deep diffusive network model with a gated diffusive unit to manage diverse information in social networks.
feature learning unit (HFLU), K Nearest Neighbors (KNN), Support Vector Machine (SVM), Naïve Bayes, Random Forest, Regression Tree, Bidirectional Encoder Representations from Transformers (BERT), GDU (Gated Diffusive Unit), Long Short Term Memory(LSTM)
1. INTRODUCTION
Soft voting classifiers in [6] incorporated four machine learning techniques i.e., Naive Bayes, Support Vector Machine (SVM), and Logistic Regression—to classify news items as false or authentic. This approach gives an accuracy of 93% for fake news classification.
As the term implies, fake news is defined as "news items that are deliberately and irrefutably false" [1] to deceive people about facts, and assertions. Particularly on social networking networks, spreading fake news has proven to be simple. Fake news is typically created to influence people's opinions, further a political agenda, or create misunderstandings about a significant topic [2]. Because of this news, elections may be impacted, ethnic conflict may increase, and criticism may be stifled.
The classification of false and true news using transfer learning on the Bidirectional Encoder Representations from Transformers (BERT) language model was done in [7]. An accuracy of 97.021 percent was achieved with this fine-tuned BERT model using NewsFN data. A Hybrid Classifier-based fake news detection system is compared with the SVM-based fake news classification system in [8]. A Hybrid classifierbased system gives more than 90% accuracy and SVM based system gives more than 84% accuracy for fake and real news classification tasks.
Fake news is of many types. They are ‘targeted misinformation, clickbait, fake headlines, fake viral posts, hoaxes, etc. Untrue piece of information spread for personal gain is known as targeted misinformation [3]. The most frequent definition of clickbait is false information created to grab readers' attention and persuade them to click on a link to a certain website. Fake headlines are the headlines that make up stuff in order to attract readers' attention and are frequently used by media and newspapers with lower credibility. Hoaxes are another sort of false information that purposefully misleads the reader by harming and financially depleting its readers.
A deep learning network named BerConvoNet has been used to classify fake and real news [11]. CNN and BERT embedding techniques are combined to create BerConvNet. In [13] CNN and Long Short-Term Memory (LSTM) have been used to classify false news items in a multimodal strategy. Here, the news was categorized according to its source and its history. This model categorizes credible news items with a training accuracy of 99.7% and validation accuracy of 97.5 %.
Various tactics to combat different kinds of false news are now being actively investigated. This paper reviews various methods of fake news recognition. The different methods are presented in Chapter
© 2022, IRJET
|
Impact Factor value: 7.529
|
ISO 9001:2008 Certified Journal
|
Page 2941