I am a PhD Candidate in the Computer Science department at Rutgers University working with Matthew Stone in Computer Science and Pernille Hemmer in Cognitive Psychology. I am mainly interested in interactive, situated, and grounded language. Grounding language is the process by which symbols in a symbol system get mapped to entities in the world. I think this is one of the most exciting frontiers of Artificial Intelligence. For the first time, natural language processing is starting to be about real things in a real way. But I am interested in more than merely mapping. I believe that, for true conversational competence, the mappings from words to worlds need to be negotiable and interactive. Humans are vague. There's no reason to expect the data we generate to be precise. We use imprecise language because it is good enough and we're ok with satisficing. To move forward, our computational models should embody this.
My research revolves around modeling speakers. Principled methods of linguistic reasoning are typically limited to toy problems. Statistical methods scale immensely well by ignoring some of the most interesting bits of linguistic theory. Using current methods from the cognitive sciences, we can scale the principled models. We can derive linguistic resources with broad coverage and the ability to license fine-grained reasoning.