DSpace Repository

Fake News Detection Using Deep Learning

Show simple item record

dc.contributor.author Limon, Foysal Ahmmed
dc.date.accessioned 2022-03-06T04:16:19Z
dc.date.available 2022-03-06T04:16:19Z
dc.date.issued 2021
dc.identifier.uri http://dspace.daffodilvarsity.edu.bd:8080/handle/123456789/7430
dc.description.abstract Fake news and misinformation have wreaked havoc on our lives in recent years. Today, fake news spreads faster and has a greater impact than ever before because of the maximum number of people who use networking as the origin of news, that happens for the prevalence of microblogs. With the rise in social media usage, it's become more important than ever to counteract the dissemination of misleading information and reduce reliance on such sites for information retrieval. Because users' interactions with fake and unreliable news contribute to its proliferation at an individual level, social networks are constantly under pressure to develop effective solutions to this problem. The public faith in the medium has been undermined as a result, having left readers puzzled. Artificial Intelligence methods for identifying false news have been the subject of extensive research. In the past, classifying online evaluations and publicly visible online social media articles received a lot of attention. In this research aims to create a model that predicts fake news, propose an optimal architecture, and then present a scientific report. This scientific paper details the most effective architecture for detecting fake news. It also aids makers of anti-fake news detection technologies in making an early choice regarding the method to take. In this study, we present a Long short-term memory (LSTM) for identifying false news. Instead of relying on custom features, our model (LSTM) employs many dense layers in a DNN (deep neural network) to extract knowledge the discriminating properties for fake news identification. Binary classifiers give prediction, cross-validation, and crips prediction at first. For improved training, time, and complexity, our model works well with this dataset. We utilize a dense layer, as do all deep learning models, to improve prediction. It works effectively and allows us to make more accurate predictions in our proposed model. We employ dropout in our model to prevent the problem of overfitting, and it works well. For recurrent neural network architecture, optimized parameters and two forms of adaptive learning algorithms were employed, in combination with which a superior outcome was picked. The proposed model was trained and evaluated using a benchmark dataset, and it provided state of the art results upon this test data, with such a 99.86% accuracy. The results were validated using several performance assessment metrics such as precision, recall, F1, accuracy, false positive, true negative, etc. These findings show considerable improvements in the identification of false news in comparison to previous state of the art results, proving the efficacy of our technique for detecting false news. The present, as well as variants of fake news, were identified using a deep learning approach. It has been observed that by combining a hybrid model with a large dataset, a better approach for detecting fake news may be proposed. Also, we didn't apply any algorithm to a dataset that was based on video or images. As we all know, these mediums may be used to promote fake news. en_US
dc.language.iso en_US en_US
dc.publisher Daffodil International University en_US
dc.subject Social media en_US
dc.subject Fake news en_US
dc.subject Deep learning en_US
dc.subject Long short-term memory en_US
dc.subject Word embedding en_US
dc.title Fake News Detection Using Deep Learning en_US
dc.type Article en_US

Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


My Account