Fake News Detection using a Modified Fully Connected Attention Mechanism for CNN BI-LSTM

Abstract

Individuals utilize social media platforms to express their perspectives and thoughts and connect with others on a wide scale. Consumers are generating more information than ever before, and sharing the information growing exponentially. Data presented on social media platforms are in an unstructured fashion, just like articles, videos, and audio. In the current digital era, the growth of falsification has triggered to the spread problem due to bogus news via social networking sites and online news sources. Some people are utilizing this helpful medium to disseminate stories that have no connection with truth. The swift spread of inaccurate information on social media platforms has become a significant problem; which brings significant negative effects to society; it might be difficult to distinguish between real news and hoaxes. Finding and identifying bogus news is challenging, manually finding fake news is tedious. Computational techniques are employed to determine whether the news data in social media platforms is authentic or not. Maintaining the integrity of information is essential. The proposed model to identify hoax news using fully modified connection attention mechanism for CNN- Bi-LSTM . It incorporates CNN and Bi-LSTM networks with a modified fully connected (Modified FC) attention mechanism. This architecture utilizes the powerful fusion of CNN with Bi-LSTM networks to collect effectively the semantic and contextual features of news articles; the Modified Fully Connected AM improves the design ability to concentrate on suitable information and detects fake news presented in social media platforms.