Programme
Title: Inference Anchoring Theory: Foundations
Katarzyna Budzynska (University of Dundee, Scotland & Polish Academy of Sciences, Poland)
Inference Anchoring Theory, IAT, combines an account of dialogue structure with argumentation theory through speech act theory. It provides an account of how dialogical actions, such as responding to a challenge, can 'anchor' structural features such as inference. This tutorial will explore the philosophical, linguistic and computational foundations of Inference Anchoring Theory. By the end of the tutorial, students will be able to apply IAT analytical techniques and understand how IAT can be used in computational domains. There are no prerequisites for the tutorial other than competence in English, and students from computational, linguistic and philosophical backgrounds are equally welcome.
Handouts, notes etc: Slides | Notes (examples for analysis)
Back to top
Title: Argumentation Mining for Educational Applications using Discourse and Diagrams
Diane Litman (University of Pittsburgh, USA)
The written and diagrammed arguments of students (and the mappings between
them) are educational data that can be automatically mined for purposes
of student assessment and instruction. Argument mining thus has the
potential to enhance many types of educational technologies including
computer-supported peer review, computerized essay grading, and massively
open online courses (MOOCs). This tutorial will focus on the use of
discourse theories and tools from computational linguistics (e.g. RST
parsers, discourse connective taggers) to support educationally-oriented
argument mining.
Handouts, notes etc: Slides
Back to top
Title: Argumentation mining: methods, challenges and possible solutions
Marie-Francine Moens (Katholieke Universiteit, Leuven, Belguim)
Argumentation mining involves automatically identifying argumentative information and its argumentative structure in text, that is, the supporting premises and conclusion of a claim, the argumentation scheme of each argument, and the argument-subargument and argument-counterargument relationships between pairs of arguments. Argumentation mining improves information retrieval and also provides the end user with instructive visualizations and summaries of the arguments. In the talk we focus on the methods to extract argumentative information, which pose interesting research questions with regard to structured machine learning. We illustrate the talk with applications that mine argumentation in legal texts, court decisions, scientific texts, debates and reviews.
Handouts, notes etc: Slides
Back to top
Title: Uncertainty in argumentation
Nir Oren (University of Aberdeen, Scotland)
This tutorial will examine the use of probability in abstract and instantiated argumentation. I will discuss the difficulties involved in reasoning with probability in argument, and describe several existing argument based approaches to reasoning with, and about, uncertainty.
Handouts, notes etc: Slides
Back to top
Title: Argumentation and mathematical proof
Alison Pease (University of Dundee, Scotland)
We will explore the relationship between argumentation and
mathematical proof and in particular consider philosophical and
linguistic differences between the processes of constructing and
presenting mathematical proof (proof-as-process and
proof-as-product). Students will do a hands-on analysis of
mathematical and non-mathematical arguments in order to determine
what, if anything, is unique in the mathematical case.
Handouts, notes etc: Slides | Tutorial sheet | Tutorial solutions
Back to top
Title: Introduction to Formal Models of Argumentation
Henry Prakken (University of Groningen & University of Utrecht, The Netherlands)
This tutorial gives an introduction to formal models of argumentation as developed in Artificial Intelligence. First Dung's famous theory of abstract argumentation frameworks is introduced. In this theory the acceptability status of arguments is defined in terms of their conflict relations while fully abstracting from their origin and content. Then an overview is given of frameworks for argumentation-based inference, in wich the structure of arguments and the nature of their conflict relations is defined and in which arguments are assumed to be constructed from a given knowledge base. Finally, dialogue systems for argumentation are reviewed, which see argumentation as a form of verbal communication between (human or artificial) agents. Both protocols for argumentation as dialogue and strategies for argumentative agent behaviour are discussed.
This tutorial presupposes introductory knowledge of standard propositional and first-order logic and elementary knowledge of set theory and the theory of relations and functions.
Handouts, notes etc: Slides
Back to top
Title: Inference Anchoring Theory: Linguistic & Technological Applications
Chris Reed (University of Dundee, Scotland)
Inference Anchoring Theory provides a coherent approach to handling a number of challenging linguistic phenomena and supports a rich range of technological developments. This tutorial will cover issues relating to ethos and persuasion in discourse, and will explore a range of computational applications. By the end of the course, students will be able to apply IAT in demanding linguistic contexts and have familiarity with a range of Argument Web applications. The only prerequisite for this tutorial is the IAT Foundations tutorial.
Back to top
Title: An introduction to the Argumentum Model of Topics
Andrea Rocci (Università della Svizzera Italiana, Switzerland)
Approaches to argument schemes are a fundamental strand of theorizing in contemporary argumentation research. Together with models of dialogue (taken broadly to include pragmatic accounts of human dialogue, informal dialectical models and rhetorical theories of status) and models of argument structure, theories of argument schemes are consider a necessary component to reconstruct natural arguments in a way that makes their soundness evaluable. Dialogue models are important to define issues and standpoints and to provide a general account of the commitment dynamics of the dialogue so that the interactional cooperativeness of the participants can be assessed. They are essential, in particular, to evaluate the quality of non-inferential moves in argumentative discussions. Models of argumentation structure offer a specialized tool to reconstruct the commitments of arguers in what concerns the relations of support between the propositions mobilized in the argument, connecting the conceded or otherwise agreed common ground to the standpoint. Models of argument schemes bring us inside each relation of support to reconstruct how an inference from the premises to the conclusion could be licensed. In the tutorial I will present a particular approach to the reconstruction of inference licenses in arguments called Argumentum Model of Topics (AMT), which was developed in Lugano by my colleagues Eddo Rigotti and Sara Greco Morasso. As suggested by the name, this approach draws extensively from a rich Ancient, Medieval and Renaissance tradition originating in Aristotle's book of the Topics, bringing this tradition back to life in a dialogue with modern views on argument schemes. One key feature of this approach is that it joins seamlessly two Aristotelian accounts of argumentative inferences that had remained hitherto separated: the syllogistic model of the (rhetorical or dialectical) enthymeme and the propositional model based on the maxims of the Topics. Thus, the AMT involves in the formal account of inferential relations both a material component where a culturally shared endoxon is brought to bear on the datum presented in the argument, and a so-called procedural component, where an abstract maxim is applied to the composition of the endoxon and datum to derive the conclusion. The second key feature of the model is represented by the conception of the maxim, that is of the inference licenses characterizing the different argument schemes. Maxims are intensional conditionals which can be seen as semantic postulates constituting the very definition of the basic relations of a (folk) ontology of the world or as theorems thereof. Efficient cause-effect, material cause-product, final cause-means, part-whole, genus-species, etc. Each of these relations, called loci in the Latin Medieval and Renaissance tradition can be seen as a node in a folk ontology of the world, defined by a bundle of semantic axioms, which represent the direct or indirect sources of the maxims employed. AMT connects the inferences drawn in ordinary arguments both with culturally specific generalizations, beliefs and values (endoxa) and with a basic ontology of the natural of social world (maxims), opening possibilities of interaction between argumentation theory and both the social and the cognitive sciences. In the first part of the tutorial the components of the model will be presented in a stepwise fashion and applied to simple examples of argument. In the second part AMT will be compared with other contemporary theories of argument schemes.
Back to top
Title: Argument Analysis in Dislog
Patrick Saint-Dizier (IRIT, Toulouse, France)
In this tutorial, we first present the Dislog language (Discourse in Logic) and how it runs on the TextCoop Platform. Concrete and simple examples will be developed.
In a second stage, we develop a method that shows how to go from argument samples in corpus to Dislog rules and lexical data. The structure of argument conclusions and supports and an annotation schema will be developed.
Finally, we show how to bind argument conclusions with their related supports, and how discourse structures may interact with arguments.
This tutorial has a practical aim: developing a small argument parser or argument mining system. The basic knowledge of Prolog that is needed will be summarized at the beginning of the tutorial. Participants may come with their own machines to test the examples.
Handouts, notes etc: Dislog tutorial
Back to top
Title: Different Approaches to Support in Argumentation Systems
Guillermo R. Simari (Universidad Nacional del Sur, Argentina)
The definition of abstract argumentation frameworks by P.M.Dung twenty years ago produced a wealth of results that have improved our understanding of the high level properties of formal argumentation. Abstract argumentation is based in considering a set whose elements are called arguments and a relation defined over that set modeling the concept of attack between arguments.
Although this simple structure has been extraordinarily fruitful, several extensions have been proposed with the aim of having better knowledge representation tools to model the argumentation process. One of the possible extensions is to introduce the notion of support in the abstract framework.
In this tutorial we will explore different facets of the support relation that have been the focus of research, such as the relations of deductive support, necessary support, evidential support, subargument, backing, and other interesting forms in which support emerges in argumentation. The goal will be to highlight similarities and differences between them, and discuss how they appear in different argumentation formalisms.
Back to top
Title: Argument Extraction from Social Media Using the General Architecture for Text Engineering (GATE) Tool
Adam Wyner (University of Aberdeen, Scotland)
The tutorial provides an introduction to argument extraction using the General Architecture for Text Engineering (GATE) tool, focussing on extraction from sample social media. GATE is a free, open-source application with a substantial development and user community. It provides a cascade of prepackaged, accessible natural language processing tools as well as the means to develop useful modules. The tutorial outlines main issues in argument extraction, shows how the tool is used, demonstrates a sample argument extraction on a small corpus, and makes several suggestions about how students can develop their own projects.
Handouts, notes etc: Slides
Back to top