Ellie {Pavlick, Tobochnik}


I am a PhD Student in the Department of Computer and Information Science at the University of Pennsylvania, working with Chris Callison-Burch. I have a BA in economics and a BM in saxophone performance (my fall back job in case this whole "information age" thing fizzles out). My main research interests are paraphasing and entailment, with some crowdsourcing on the side.

Paraphrasing and Entailment: I am interested in paraphrasing, and particularly in times when the loose definition of "meaning pretty much the same thing" isn't good enough. I am looking at the entailment properties of paraphrases-- e.g. whether they are symmetric (twelve/12) or directed (girl/person). I am working on using lightweight formalisms like natural logic to predict how entailment relations between words can be composed to infer entailment relations between phrases. I am also working on how context-- e.g. domain, style, and local dependencies-- affects the goodness of a paraphrase.

Crowdsourcing: I am involved in a variety of projects and collaborations involving crowdsourcing. I have worked with using Amazon Mechanical Turk to gather data for machine translation, recognizing textual entailment (RTE), and other language understanding tasks. I am also interested in crowdsourcing for social science applications-- right now, I am particularly excited about our current project using crowdsourcing to build a database of gun violence in the United States, which we are building as part of the undergraduate course that I co-developed with Chris.