The Question-Answering problem has seen a rapidly increasing interest in the past few years. Despite the surge of proposals, research has shown existing methods possess a very shallow understanding of text, which is reflected in their poor generalization capabilities. A possible reason for this behavior is these systems have been trying to address MC as a yet-another-ML problem, without focussing on the diverse underlying challenges, which might involve a deep understanding of not just linguistic phenomena and ML techniques but also Knowledge Representation, Logic, Reasoning, etc.
In this talk, I will approach the Question Answering problem, from the “reasoning” perspective. I will go over the state of the field and its current limitations. I will introduce a formulation for abductive reasoning in natural language and show its effectiveness in two different domains of elementary school science exams and biology tests. Next, I will introduce MultiRC, a reading-comprehension playground where questions can only be answered based on information from multiple sentences.
Daniel Khashabi is PhD candidate with Professor Dan Roth at the University of Pennsylvania. His interests lie at the intersection of computational intelligence and natural language processing. Daniel obtained his B.Sc. from Amirkabir University of Technology (Tehran Polytechnic) in 2012, and spent a few years as a graduate researcher at University of Illinois, Urbana-Champaign, before moving to UPenn in 2017.