Colloquia / Seminars / Industry Talks

Natural language processing (NLP) is often said to be "AI-complete", meaning that the most challenging problems in artificial intelligence manifest themselves in natural language phenomena. Some researchers observed that the majority of the NLP tasks follow the 80/20 rule: while 80% of the task can be addressed using shallow text understanding (the "easy victories"), the remaining 20% can only be solved by reasoning with sophisticated knowledge (the "uphill battles"). Now that the easy victories have largely been achieved, NLP researchers have started the uphill battles in recent years.

I will begin this talk with an overview of the ongoing projects in my lab, which belong to four areas of language understanding, namely, automated essay grading, argumentation mining, coreference resolution, and joint inference for NLP. I will describe the victories we have achieved and the battles we are currently facing, sketching the long-term goals and vision we have for each project.

The second part of the talk will focus on two tasks in two of these areas. The first task is a pronoun resolution task known as the Winograd Schema Challenge, which has recently been proposed as an appealing alternative to the Turing Test. The second task involves extracting complex biomedical events from text. For each task, I will describe our group's attempt to fight the battles by reasoning with background knowledge.

Vincent Ng is an Associate Professor in the Computer Science Department at the University of Texas at Dallas. He obtained his B.S.in Computer Science from Carnegie Mellon University and a Ph.D. in Computer Science from Cornell University. His primary area of research is machine learning for natural language processing. He is/was on the editorial board of the Computational Linguistics journal and ACM Transactions of the Asian and Low-Resource Language Information Processing, and is a guest editor of a special issue of the Natural Language Engineering journal. He has served as an area chair/senior program committee member for conferences such as AAAI, ACL, NAACL and COLING, and has organized workshops and given tutorials on a variety of topics in natural language understanding in both AI and NLP conferences.