Fake News Detection using Machine Learning – Present and Future Trends

The proliferation of yellow journalism or fake news and the way it is spreading especially on social media has become a big concern because of its devastating effects. At this stage when the information we require is under our fingertips, on the web we see a lot of cooked-up stories designed to make people believe something. This may involve buying a particular product, visit a website, or even sensitive information about religion, community, etc. This is nothing but 'fake news which is a type of yellow journalism that consists of misinformation or hoaxes spread via online social media or even on print and broadcast news channels. There's almost no restriction by social media platforms for anyone to publish his/her thoughts. In such platforms there's a big issue, that most of the people do not verify the source of the information that they browse online before they share it, thus leading to fake news spreading rapidly or even ''going viral. Moreover, it's very difficult to identify the source of such misinformation thus making it harder to assess their accuracy. The result of fake news is that people try to believe in excuses, reject others' ideas, avoid the truth, and spread rumors. It can harm all segments of society especially in workplaces where people are cynical and unsure of who to believe. Social media has become a dominant source of news and information and has dramatically reshaped these industries. However, fake news existed long before the arrival of social media. It became a buzzword after the US presidential elections in 2016. The internet has given a boom to fake news, regardless of how the information is misinterpreted: whether it is a rumor, fake stories, or fallacious reporting.

The good news is that in near future artificial intelligence or to be more specific machine learning-based models will help a user to check whether the news is real or fake. Even though the research in this area is going on, but the researchers claim that there's still a lot more to be solved. This particular area of research comes under the hierarchy of machine learning known as natural language processing (NLP). This area is receiving a lot of attention from researchers, scholars, and academicians across the globe. The number of users of online media is growing therefore automated detection of fake news seems to be the only way to tackle such a problem. So far there have been text-based detection approaches of fake news which did not yield better results. Almost all the machine learning models use hand-crafted features extracted from input textual content. In the future, we will witness a context-based approach in detecting fake news. In 2016, some researchers found that the traffic taken by Facebook is almost 50 percent fake and hyperpartisan, while at the same time news agencies depend on Facebook for 20 percent of their traffic.

Fake news has been spreading through Twitter also. Recently it was found that fake news being tweeted during the COVID-19 pandemic for the purpose to mislead the targeted population. This has also become a basis for new research that exhibits a new approach to detect fake news on Twitter. Machine learning, as well as highly sophisticated deep learning models, are being continuously used by researchers and industrialists to developed automated fake news detection-based models. Many such models detect news of particular types such as political and religion-based. Some research journals reveal that such models have features for specific datasets that match their topic of interest. Such approaches might suffer from dataset bias and perform poorly on news of another topic. Deep learning-based models are bringing a revolution in almost every walk of life. The recent developments in this field in natural language processing tasks, make them a promising solution for fake news detection. With the advent of Keras (an API in deep learning) and Tensorflow (end-to-end open-source platform for machine learning), coding and implementation of such intelligent models have become a lot easier compared to a decade back. The future will witness deep learning models to have a great prospect in fake news detection. Such models will be able to classify between fake news and legitimate news. Decades of deception detection have shown how well we humans can detect lies in the text. The findings show that we are not so good at it. In reality just 4 percent better than chance, based on a meta-analysis of more than 200 experiments. The viral spread of fake news hurts the behaviors, beliefs, and attitudes of the public which can seriously endanger the democratic processes. Early detection of such false information and extensive spread presents the main challenge for researchers today.

Dr. Akhter Mohiuddin Rather

Associate professor,

Great lakes Institute of Management, Gurgaon