Tool to analyse mental state of social media users

Everyday millions of people around the world post numerous messages and comments on various social networks. What if there was a tool that could analyse a collection of these posts and indicate a person’s state of mind?

The idea to help people who are depressed and feeling low was something that was very dear to Kumar Nitya Suman, a third year information science student of PESIT, Electronic City. At the 24-hour hackathon organised by Microsoft India recently, Suman brought this idea to its logical conclusion by developing a tool which in his words, could “understand the mental condition of users in social networking sites and try help those in need.” Suman and his team came third in the competition.

To test his tool, Suman used profiles of over 100 Twitter users from well-known figures such as cricketer Sachin Tendulkar, Microsoft CEO Satya Nadella and commoners such as his own brother and friends. The tool has features such as text analysis of keywords, a panel that manages tweets and sends out notifications, a scoring system that gauges a person’s mood - from low to high - based on analyses.

“A person can add someone close who can see the mood score and  therefore monitor him or her. Privacy therefore remains. The algorithm is such that one can see the decreasing or increasing moods and notifications are sent accordingly,” said Suman

The hackathon invited students and early-stage developers to build working applications with the help of various tools such as Microsoft Cognitive Services, Microsoft Bot Framework and Microsoft R Services.

The inspiration for the “sentiment analysis tool” came when Suman’s classmate in college vanished from the institute to pursue his passion of singing. “All this while I did not know what he really wanted. What was on his mind. He wanted to become a singer but pressures of doing something else did him in,” he said. Further, Suman said that with about 1.7 billion people on Facebook and 500 million on Twitter, there would be a way for a sentiment analysis and "how people behaved through their activity in such social networks” The tool can be further developed as a web application for search engines such as Google or social networks like Facebook, Twitter or a standalone phone application or a website. “It can definitely be scaled up into any of these form,” he said.