Whether WhatsApp should consider machine learning technology to combat the fake news threat in India? This was the thought that fascinated a researcher from IIT-Delhi. The researcher Ponnurangam Kumaraguru, fondly addressed as PK, is currently working on a proof-of-concept. WhatsApp is under vigil in India for failing to prevent the spread of fake news and rumors, that has resulted in the outbreak of several dreadful crimes in the country. While the instant messaging platform is taking steps to address this challenge, machine learning-based experiments are being conducted to fix the problem. PK says that their upcoming machine learning-based solution is capable of identifying fake text, images, videos and audio messages.

How WhatsApp works?

According to a recent report, the application will segregate fake news and genuine information by using color codes or labels. For instance, a message labeled green indicates that the message is genuine, while a red label indicates that the message is bogus. Ideally, the application will work as an extension or wrap around the principal WhatsApp application.

Currently, the research team is collecting data sets from users. The data collected will be processed to feed the necessary information into the application. The application will do annotations of the posts. That means, the machine learning model will identify which feature or element of the fake news is stronger and which is not. When you use the application on your phone, it will look at each of your messages and analyze them through machine learning, giving annotations for each post. Thus, you will come to know about the genuineness of the message. The app will also rate a message by providing a credibility score.

Privacy Concerns

The app will have to read users’ messages in order to support content flagging. WhatsApp offers end-to-end encryption on its platform that blocks a third-party (other than the sender or receiver) from accessing the messages. However, privacy might not be a major concern for this machine learning-based app, because the application is designed to work offline and on-device. Since the data is not sent to any of the servers and the application works on-device, it will not violate user privacy, explains the IIT researcher.

Recently, WhatsApp has brought in restrictions with respect to message forwarding in India, in order to prevent the proliferation of fake and provocative messages. Platforms that combine complex human intelligence and the right technology to curb such issues are the need of the hour.

Related read: Five Reasons Your Business Should Adopt Machine Learning.