Finding support for India during its COVID-19 surge
India and Pakistan have waged four wars in the past decades, but when India faced a shortage of oxygen in its hospitals during its recent surge of COVID-19, Pakistan offered to help.
On Twitter, hashtags like #IndiaNeedsOxygen and #PakistanStandsWithIndia were all the rage. However, finding those positive tweets wasn’t as easy as just browsing the supporting hashtags or checking out the most popular posts. Negative tweets often hijack supporting hashtags to troll or fight with other users. And Twitter’s algorithm isn’t tuned to bring up the most positive tweets during a crisis.
Ashique KhudaBukhsh from the Institute of Language Technologies at Carnegie Mellon University led a team of researchers who used machine learning to identify supportive tweets from Pakistan during India’s COVID crisis. In the midst of a public health crisis, words of hope can be welcome remedies.
“When a country experiences a national health crisis, the more messages of support you see, the better,” KhudaBukhsh said. “If you do a random search, you will find positive tweets about 44% of the time. Our method gives a person positive tweets 83% of the time. That person will have to do a lot less work to find the supporting tweets.”
By combining existing language classifiers – algorithms formed by machine learning that determine, for example, whether the speech is positive or negative, hopeful or distressing – the team developed a system that identified Pakistani tweets. favorable or positive on India in 83% much better than the existing methods. The team used their method to conclude that more than 85% of tweets posted on India’s COVID crisis from Pakistan were favorable.
“We have limits but not in our hearts,” began a tweet detected by the team.
The team included KhudaBukhsh, Clay H. Yoo and Rupak Sarkar from LTI and Shriphani Palakodety, an engineer from blockchain and AI company Onai who earned his masters degree from LTI. They published their findings in an article titled “Empathy and Hope: Resource Transfer To Model Inter-country Social Media Dynamics,” which was accepted into the ACL Association for Computational Linguists’ Workshop on Natural Language Processing for a. positive impact. The work was done in real time with the crisis and as team members worried about the health of loved ones in India.
Research is important in several ways. First, the team showed that existing language classifiers can be useful in large contexts. This is important because to deploy a classifier that will be useful in the midst of a crisis, it must be built quickly. It can’t be built from scratch, and the team wanted to see if existing research on language classifiers could help.
To detect supportive tweets during India’s COVID crisis, the team used a hopeful speech classifier that KhudaBukhsh and Palakodety built with the late Jaime Carbonell, a prominent professor at the School of Computer Science who founded LTI. , to identify positive YouTube comments on videos posted about the 2019 escalation of the India-Pakistan conflict over Kashmir. The team then combined the hope-speech classifier with a known empathy-distress classifier.
Although these two language classifiers are built for different reasons and trained on different data, they effectively detected positive tweets during the COVID outbreak in India.
“We have shown that there is a kind of universality in the way we express our emotions,” KhudaBukhsh said. “And we have shown that we can use existing solutions, combine them and quickly tackle future crises.
The research was also potentially important for the crisis in India. KhudaBukhsh and Carbonell envisioned the Hope Speech Classifier as an alternative way to tackle hate speech. Instead of detecting and suppressing, minimizing or blocking hate speech – which exists en masse on the internet – the two sought to use their hopeful speech classifier to identify and amplify messages of support. People are influenced by what they see and read, and if messages of hope are placed in front of them rather than hate messages, it could affect the way they think and act.
The team identified tweets that offered prayers to India, spoke to the common humanity of two countries, and sent love.
“It is heartbreaking to see this situation in our neighborhood. Send love and prayers from Pakistan. May Allah Almighty help mankind through this pandemic. Stay strong. Stay safe,” reads. on in a tweet found by the team.
Focusing on support between India and Pakistan could make the difference, KhudaBukhsh said. And since so much fighting is now taking place on the internet, maybe this is the place to start.
“These two countries have such an acrimonious past,” KhudaBukhsh said. “Any positive behavior on either side can help promote world peace.”
Warning: AAAS and EurekAlert! are not responsible for the accuracy of any press releases posted on EurekAlert! by contributing institutions or for the use of any information via the EurekAlert system.