Breaking New Ground in Machine Learning

Sharing Innovative Work

Interested in how machine learning can be used to measure sarcasm on social media? Then step forward Aniruddha Ghosh. A PhD student at UCD, Ani is an active member of Dublin’s Machine Learning Meetup Group – and spoke at the recent meetup in Bank of Ireland Grand Canal. Currently using machine learning to help identify what qualifies as sarcasm on Twitter, Ani says of his work: “Right now I’m working to identify sarcasm and then later on, I want to work to generate sarcasm as well. It’s still a work in progress, we have to do a lot of work on generating models using deep learning but it’s really interesting.” Now in his third year of work on the project, Ani is excited about the possibilities that the project offers.

It’s still a work in progress, we have to do a lot of work on generating models using deep learning but it’s really interesting.

Using Machine Learning

He says: “My priority now is to identify how sarcastic the language (on social media) is. It’s not easy for people to understand. Sarcasm is subjective – so some people think it’s rude and some people think it’s funny. I work to identify what actually qualifies as sarcasm, so it will be easy for people to understand.”

“We always want to make our system capable of identifying what is sarcastic or non-sarcastic. We need to treat it like it’s a function of probabilities. I want to create a system that can identify and tell people how sarcastic some language on social media is,” he adds. “I started using deep learning last October and then built the system.”

Looking to the future, Ani is keen for the project to continue to develop. He jokes: “I’m building the system and then I’ll see who wants to buy it! If someone is interested and wants to take the work further in the future – why not? It’s about the research and work. I love language and I love working with computers. It’s a good mix, we learn about sentiment and how it’s applied.”

Plans for the Future

With one more paper due to complete his PhD, Ani is keen to continue working in this field. He says: “Dublin is really cool, my supervisor is awesome to work with, and the entire team at Adapt are great. I get my data and share it with them and they annotated it in one week. That’s really helped a lot. After completing my PhD, if I get an opportunity over here, I would stay as the research environment and tech scene in Ireland is really good.”

Along with Ani, John Kelleher also spoke at last month’s meetup on An Introduction to Regularisation. A popular monthly event, anyone interested in attending the next meetup can check out the Machine Learning Dublin’s meetup page and our events section.