Comparing similarity between the title and the content of news articles using deep learning

Abstract

The volume of the online news has rapidly increased for recent years. People who consume the news online have increased. New technologies have changed the way of consuming online news. It has become a huge adoption of accessing news with the help of smart devices. When people read online news, they scan the headlines first and then pay attention to the content. Online Readers visit the articles in brief based on the headlines. Spotting fake news from factual news is a challenging task. This paper presents a method to separate fake news from factual news by comparing the similarities between the headlines and the content of the articles using deep learning methods. In this proposed method, the abstractive text summarization technique is used to extract the summary and then generate new headlines for the articles. The selected dataset has 60 000 news articles and collected from a benchmark dataset, BBC news. This model performs the best effort and achieves the accuracy 74% for selected samples.