Computational Argumentation: pragmatics and problem structuring

Public Event
21 October 2016
Begin time: 

In this talk, Simone Teufel will discuss two areas in computational argumentation that she is currently involved in. The first concerns the task of enthymeme reconstruction - the task of inserting missing premises into an argument. The task can be seen as a reverse entailment detection task, and address it using pragmatics and argumentation schemas such as Walton's, and pragmatic inference. Teufel looked at mini-arguments using the cue phrases such as "therefore", "of course", "let alone" and "because". Teufel will report results of a feasibility study and discuss evaluation. As for the second area of my talk, it concerns scientific argumentation and the role of problems, tasks and solutions in it. This is an attempt to improve my long-standing research in recognising argumentation and rhetorics in scientific articles in a cross-disciplinary way. In a standard supervised ML framework, we consider which features (from inside the potential description of a problem, versus from its context) to use to identify problems. Simone Teufel will discuss how to extend this approach to finding solutions, and talk in particularly about the connection between the two. 

Simone Teufel is a Reader in Information and Language in the Natural Language and Information Processing Group at Cambridge University's Computer Laboratory. Her PhD research (1999, Edinburgh University) looked at the application of discourse learning for summarisation; her first degree, in Computer Science, is from Stuttgart University. During a Postdoc position at Columbia University in 2000-2001, she was also working on medical information retrieval. She is particularly interested in models of human argumentation, text understanding and summarisation which are cognitively explanatory. Her other research interests include large-scale, applied NLP, information retrieval, summarisation evaluation and discourse parsing.