Who Shot the Elephant in Their Pajamas? Commonsense Reasoning for Natural Language Processing Systems

This seminar is CSCI 6999 eligible


Jackie C. K. Cheung, McGill University


How do we know that elephants do not generally wear pajamas? Knowledge about possible, plausible, and likely events-or the lack thereof-is one of the main bottlenecks in developing intelligent natural language processing systems capable of fluid interactions with human users. In this talk, I will present our work on discovering and using such knowledge from the web for so-called commonsense reasoning tasks. I will describe an approach based on information retrieval applied to the entire indexed web to tackle the Winograd Schema Challenge (WSC), a difficult commonsense reasoning task that involves resolving pronominal anaphora. I will also discuss challenges in the evaluation of commonsense reasoning, examining claims about recent improvements on WSC benchmark results. Finally, I will describe our new KnowRef corpus, a set of pronoun disambiguation questions inspired by WSC which is an order of magnitude larger yet maintains much of the difficulty of WSC.


Jackie Chi Kit Cheung is an Assistant Professor at McGill University's School of Computer Science, where he co-directs the Reasoning and Learning Lab, and a Canada CIFAR AI Chair at the Mila Research Institute. He obtained his PhD from the University of Toronto. His research focuses on developing computational methods for understanding text and for generating language that is fluent and useful in context. Dr. Cheung is an academic advisor for the Borealis AI research lab in Montreal. He was a Program Co-Chair of Canadian AI 2018, and received a best paper award at ACL 2018.




Slonim Conference Room (#430)
Goldberg Computer Science Building
Dalhousie University
6050 University Avenue, Halifax