Speaker:Lu Wang
Abstract: Understanding, evaluating, and generating arguments are crucial elements of the decision-making and reasoning process. A multitude of arguments and counter-arguments are constructed on a daily basis to persuade and inform us on a wide range of issues. However, constructing persuasive arguments is a challenging task for both human and computers, as it requires credible evidence, rigorous logical reasoning, and sometimes emotional appeals.
In this talk, I will introduce our neural network-based argument generation model. It consists of a powerful retrieval system and a novel two-step generation model, where a text planning decoder first decides on the main talking points and a proper language style for each sentence, then a content realization component constructs an informative and fluent paragraph-level argument. We believe that the proposed argument generation framework will enable many compelling applications, including providing unbiased perspectives on complex issues, debate coaching, and essay writing tutoring. Our framework is also generic and has been applied to other text generation problems, such as Wikipedia article paragraph generation and scientific paper abstract writing.
Bio: Lu Wang is an Assistant Professor in the College of Computer and Information Science at Northeastern University. She earned her PhD in Computer Science from Cornell University, her BS in Intelligence Science and Engineering and her B.Econ in Economics from Peking University. Professor Wang is interested in developing natural language processing and machine learning techniques to help people efficiently and effectively understand and absorb knowledge from large-scale text data with inherent noise. She received an outstanding short paper award at ACL 2017 and a best paper nomination award at SIGDIAL 2012.