On November 30, the Center for Data Science hosted NYU’s first Natural Language Processing (NLP) Reception. Led by esteemed professors Sam Bowman and Kyunghyun Cho, and sponsored by the Moore Sloan Data Science partnership and CDS, researchers working with text-as-data came to network with like-minded scholars, present their projects, and exchange ideas at the reception.
Both Bowman and Cho have produced groundbreaking NLP research that has helped to shape their field. Presently, Bowman is investigating how unsupervised deep learning can help sentence analysis, while Cho is examining how neural networks can improve computer-generated translations like Google Translate. But alongside their innovative work, the reception featured ten scholars from Engineering, Law, Marketing, and the Social Sciences who are also making fascinating contributions to the field.
Assistant Professor Xiao Liu from Marketing, for example, is interested in using unsupervised deep learning to investigate how online reviews can affect sales on platforms like Amazon. Which reviews are being read, and how much do they influence the customer?
Kevin Munger from Politics, however, is using NLP to reduce political incivility on platforms like Twitter. By using an algorithm to detect aggression, he intends to create robot Twitter accounts that will intervene in various ways to restore equity online.
Like Munger, Dr. Rumi Chunara with her research assistant, Kunal Relia from Computer Science and Engineering, are also working on Twitter, but with a different emphasis. After gathering a dataset of almost 18 million tweets that were produced in New York, they used Amazon Mechanical Turk to sort and label the tweets that were homophobic or racist. Sorting these tweets allows them to identify specific neighborhoods where homophobia and racism are most prevalent in the city.
Masatoshi Tsuchiya, a visiting researcher from Japan’s Toyohashi University hopes to use NLP to improve cross-lingual translations and generate automatic summaries. Improving these processes, Tsuchiya believes, will help those who are learning English is a second or third language in Japan and elsewhere.
Collaboration has always been central to the interdisciplinary spirit of CDS, and it is heartwarming to witness the rapid exchange of ideas between such a diverse range of scholars.
“I’ve wanted to build an NLP community at NYU since I arrived. Part of my job is finding ways to open up a greater dialogue between the social sciences and the data sciences, like hosting this event. These events are also an opportunity for professors to meet graduate students, who are really efficient resources for collaborating on large research projects. I’m hoping to do smaller, more regular gatherings in the future, and then perhaps another university-wide reception again in a year or two,” Bowman said at the end of the night.
by Cherrie Kwok