Text Analytics: Natural Language Argumentation

Text Analytics: Natural Language Argumentation

Course Description

Arguing is a fundamental aspect of human communication and reasoning. We may argue with our friends about which movie to watch, with experts about the long-term effects of global warming, or with ourselves about which of various job offers to take. But what exactly is an argument? What components is it made of? What is the interplay between different arguments? And what makes an argument convincing?

In recent years, Natural Language Processing (NLP) and Machine Learning methods have been applied to try and answer these question, resulting in a new research area called “Argument Mining”. This seminar will introduce some theoretical ideas and assumptions underlying many of the works in Argument Mining and review in depth the latest research in the field. Many of them are based on state-of-the-art Machine Learning methods.

Topics covered in the seminar are, among others, the automatic detection of claims and evidence in text, the identification of contradictory and supporting arguments in debates, how to predict the convincingness of arguments, and how to judge if two arguments are making the same point. We will also discuss exciting applications of Argument Mining research, including detecting deceptive reviews, predicting the winner of a debate, and automatically grading essays.

Organisation

Lecture: Thursday 13:30-15:10, Room S105/22

The first class will be held on October 20th 2016.

Additional material will be distributed via the Moodle eLearning platform. The required passcode will be announced during the first lecture.

Teaching Staff

  • Dr. Edwin Simpson
  • Prof. Dr. Iryna Gurevych

We do not have fixed office hours. Please register via email if you need an appointment.

Literature

Will be announced during the seminar.

Timetable

The first sessions will feature introductory lectures on different ideas of what constitutes an argument, early Argument Mining approaches, and relevant machine learning and NLP concepts. The program for the remainder of the seminar will be determined according to the number of participants and will cover the following topics (not necessarily in this order):

  • Task design and user interfaces
  • Incentive mechanisms and gamification
  • Quality control and aggregation
  • Task allocation and active learning
  • Budget optimisation
  • NLP Applications
  • New directions in crowdsourcing + machine intelligence