Fall 2019 Natural Language Processing - Matt Gardner

Explore a thought-provoking lecture on machine reading comprehension and the challenges of assessing true machine understanding. Delve into the current state of machine performance on reading comprehension tasks and the skepticism surrounding claims of human-level ability. Examine the key elements necessary to create a comprehensive benchmark for evaluating genuine machine reading comprehension. Learn about the speaker's proposed approach to developing such a benchmark and the initial progress made in this direction. Gain insights from Matt Gardner, a senior research scientist at the Allen Institute for AI, as he shares his expertise in natural language processing, question answering, and semantic parsing. Discover the intersection of open domain reading comprehension and formal grounding in question semantics. Understand the importance of reasoning over open domain text in advancing machine reading capabilities.