Telengana can mine public posts on social media

Hyderabad: The government can observe social media conversations of people if privacy settings are in public mode. Social media companies such as Facebook and Twitter sell the raw data if it is in public domain. Since large amount of data is available over the social media, it can be accessed using a suitable software. Furthermore, such data can be used for policy matters. Additional DG of PIB, B. Narayanan, spoke on ‘using social media to monitor citizen sentiments for various initiatives of the government as well as to take customer feedback.’ 

He said that now information has to be posted on social media like Facebook and Twitter as social media is where the youth are. “People search through these websites and come out with sentiments on topics. Softwares can be used to access public’s mood from this data,” he said. If the government wants to know the mood of the people on Swachh Bharat campaign, the software scans social media to get the data and it will assess the mood. 

“Off-the shelf software by Oracle, Meltwater, One direct, Radian six etc are customisable and can mine data and search through social media posts of people using suitable key words and a rudimentary artificiaI intelligence, ‘mine’ trends, sentiment and feedback. It gives a sense on the favourability of a campaign,”he said. 

However, reports extracted are often sent to the respective government departments who can use it or ignore it. However, much of the social media chatter is celebrity oriented, but given large volumes on some issues, it can give a reasonable accuracy of feedback. 

“Using natural language processing, machine learning, ‘topics can be created.’ Furthermore, data from past dates can be searched. Due to the exorbitant prices charged by the platform, it is cheaper to buy it from aggregators ie software companies,” he said. Due to increase in organised trolling, where in fake accounts and Bots are used, they tend to pollute the samples and disrupt the accuracy of the feedback. 

This is because Artificial Intelligence (AI) cannot differentiate if the Tweet is being put up by a real person or a fake ID. Meanwhile, the feedback is still limited to Indian languages as the software cannot scan in local languages. “Even though social media platforms have opened interfaces in 8 or 9 languages, unlike English, there is paucity of information in other languages,” Narayanan said.