TT Epaper
The Telegraph
Graphiti
 
CIMA Gallary

Tighter tracker for social media

New Delhi, July 16: The government is examining proposals from the information technology industry to intensify real-time analysis of social media platforms for keeping tabs on developing events, public mood and impending threats.

Sources said senior officials of the National Security Council held a meeting with representatives of software giant Infosys late last month after which a committee was constituted to brief the private company for developing a web application portal that can automatically search and scrape data off social networking sites based on specific queries on keywords.

The panel, headed by Raghav Bhatnagar, joint secretary in the cabinet secretariat, also has officials from the information and broadcasting ministry and is being supervised by the security agency. The committee has already held a meeting with the Infosys team earlier this month.

On a previous occasion, another major IT firm, Wipro, had come up with a similar representation, but the government is yet to assign the project to either of these firms.

The move is in line with Prime Minister Narendra Modi’s keen interest in using social media tools to gauge public mood.

“At present, the I&B ministry prepares a daily report of trends and comments to get a pulse of the nation by keeping a watch on social networking sites such as Twitter, Facebook and blogs. There is a team of professionals in the ministry who do content and trend analysis of issues that are being talked about by people on social media platforms. However, the government wants to be more specific and systematic now and is set to gather intelligence on social media,” said a source.

In the US, the Federal Bureau of Investigation had also sought industry assistance about two years ago for a social media monitoring tool, much to the discomfort of privacy advocates.

While panel members contacted by The Telegraph declined to comment, computer science and data-mining experts said the volume and velocity of data streaming into social media networks makes it impossible for human analysts to meaningfully screen the content. Real-time tracking of social media networks is best done through special software to detect “events” — or bursts of chatter.

“Such a system could be used to assess public mood, for example to seek feedback on government schemes, find out whether a government initiative is popular or unpopular,” said Dheeraj Sanghi, a professor of computer science at the Indian Institute of Technology, Kanpur.

It could also be used to track, Sanghi said, sensitive chatter. “Someone may make derogatory comments, or some event somewhere may stir public emotions — during such situations, the government could use an event tracking system as an early warning system,” he said.

Experts say social media platforms could also be used to track and stop waves of misinformation. “The advantage of using social media instead of other channels is that misinformation can be detected and stopped at very early stages,” said Daniel Romero, a mathematician who specialises in social and information networks at the University of Michigan in the US.

The rise of software tools to track social media content has raised a debate over the issue of breaches of privacy with some people concerned that their views expressed on social media sites may be misinterpreted or portrayed out of context.

“I think the real concern is that many people are not aware of just how public the content they publish on social media sites really is,” Romero told The Telegraph. “People need to be aware that not only is their content public, but that algorithms can track very specific information. If the right expectation of privacy is clear for everyone, the interactions between citizens and governments on social media would be less perceived by citizens as the government spying on them.”

The key to event detection is to find relevant activities in the huge volumes of data. “There is a lot of noise in social media networks,” said Abhishek Roy, a computer scientist at the University of Massachusetts, who has worked on event extraction from a public database of photographs.

More than 500 million tweets are posted a day. Most of these tweets are likely to be trivial. “The primary objective of event detection on social media is to find relevant events in the huge influx of tweets,” Roy said.