Computers can sift through large volumes of data, analyse it and come out with actionable insights. But what it lacks is a funny bone. Though efforts are on by computational researchers across the world to make computers understand humour, not much research is happening in Indian languages.

Researchers at the International Institute of Information Technology (IIIT-Hyderabad) are working on a project, classifying humour and letting the computers understand it and generate humour. To begin with, they working on Hindi and Telugu texts.

“When you read a few sentences in a book, they evoke a kind of emotion in you. But for the computer, they are just words. At best, it can understand grammar. The challenge is to make it understand the context and humour,” Radhika Mamidi, Associate Professor at the IIIT-H, said.

As part of her PhD work in the early 1990s, she analysed Kanyasulkam (Bride Price), a satirical play penned by Gurazada Appa Rao a century ago.

She is now using the same text to classify humour to make it easier for the machines to learn.

She and her students Srishti Aggarwal, Vikram Ahuja, Taradheesh, Vaishnavi, Gayatri have been involved in Humour Analysis and Generation.

Vaishnavi is working on the conversational humour in Kanyasulkam to help machines capture cultural nuances of such humour.

Covid humour

Radhika’s team also studied how people fell back on humour to cope with extreme stress induced by prolonged lockdowns during the Covid-19 pandemic.

“You are caught in a crisis, confined to homes not knowing how to tackle the crisis. Humour can be a good vent, letting people laugh at their own misfortune,” she said.

“We have started collecting jokes, memes, videos and GIFs in English and in Telugu on the pandemic,” she said.

Gayatri Purgilla, a Dual-Degree student at the institute, tracked and analysed what kind of jokes or videos went viral thru the pandemic.

Problems triggered by the pandemic (shortage of toilet paper, for one) led to creating a torrent of jokes.