IIIT-Delhi scans social media use with AI to predict personality
Working with Mahindra Ecole Centrale of Hyderabad, they have developed the system using machine learning, natural language processing and social network analysis. This, they say, analyses content posted by an individual on social media, including Twitter, based on 70 categorical features.
“This system helps in understanding the functioning of society and provides wide possible applications, such as internet advertisement, poll analysis, understanding depression, counter-terrorism etc,” said Tanmoy Chakraborty, faculty member at IIIT-D and the lead researcher on this project. “This was a paid exercise. Once we gathered enough data, we started building our AI models.”
Data from over 32,000 users was collected from the top 50 cities, with at least 500 users being selected randomly from each city. “We chose only those profiles that posted more than 100 tweets and were at least one year old,” said Chakraborty.
IIIT-Delhi and Mahindra Ecole are currently working with researchers from the Post Graduate Institute of Medical Education and Research (PGIMER), Chandigarh, to understand the status of general mental depression in India.
The model is also being used by a leading software services company for its recruitment process, he said.
“This research can (also) be used for political campaigns, but we are sceptical and progressing cautiously as we do not want to repeat the story of Cambridge Analytica,” Chakraborty said.