Scottish Argumentation Day 2013
The 2013 Scottish Argumentation Day will be hosted by ARG:dundee on Friday 19th July. The main aim of the day is to allow researchers in argumentation and related fields from across Scotland to come together and present their work in an informal setting. The day follows on from the successful first meeting held in Aberdeen in 2011.
The venue for the event will be the Wolfson Research Theatre in the Queen Mother Building (where the School of Computing is based, building 26 on the campus map).
When you enter the building, go through the double doors on your right and it’s just past the lift.
The provisional schedule is as follows. Titles and abstracts for the talks will be provided in due course:
- 09.30-09.55: Arrival
- 09.55-10.00: Welcome
- 10.00-11.20: Session 1
- 10.00: Nir Oren
Explaining automated reasoning via argumentation (Abstract)
- 10.20: Adam Wyner
On the Instantiation of Knowledge Bases in Abstract Argumentation Frameworks (Abstract)
- 10.40: Mark Snaith
Argument Revision as a means of supporting dishonesty (Abstract)
- 11.00: Martin Caminada
On the Equivalence between Abstract Argumentation and Logic (Abstract)
- 10.00: Nir Oren
- 11.20-11.40: Tea/coffee break
- 11.40-12.40: Session 2
- 12.40-14.00: Lunch
- With the option of a visit to the ARG:dundee Lab to see a demonstration of the Argument AnalysisWall
- 14.00-15.20: Session 3
- 14.00: Hengfei Li
Probabilistic Argumentation (Abstract)
- 14.20: Rolando Medellin
Interaction with broadcast debates: The Grumpy Old Debate project (Abstract)
- 14.40: Federico Cerutti
Argumentation Theory and Description Logic: an Engineering
- 15.00: Simon Wells
Aligning Argumentation Theory with Behaviour Change Mechanisms (Abstract)
- 14.00: Hengfei Li
- 15.20-15.40: Tea/coffee break
- 15.40-16.50: Open discussion on future directions
- 16.50-17.00: Closing remarks
Explaining automated reasoning via argumentation
Nir Oren, University of Aberdeen
In this short talk I will examine the use of argumentation to explain complex concepts to non-expert users. An exemplar of this approach is the EPSRC funded SAsSy project, where we aim to make distributed autonomous systems scrutable tp users through the use of argumentation and natural language generation techniques. More specifically, by engaging in dialogue, we seek to enable the user to understand why some plan was selected by the system, to understand the consequences of the plan, and to be able to modify the plan as needed.
Argument Revision as a means of supporting dishonesty
Mark Snaith, University of Dundee
This talk will start by showing that dishonesty can be characterised as a form of internal Belief Revision and, by extension, Argument Revision. Then, using the dishonesty typology characterised by Sakama et. al. (2010), it will be demonstrated that applying Argument Revision techniques can assist an agent in being dishonest in a dialogue in a 3-stage process: 1) deciding whether or not to be dishonest; 2) which dishonest act to perform; and 3) supporting the dishonesty to avoid detection.
On the Equivalence between Abstract Argumentation and Logic
Martin Caminada, University of Aberdeen
Abstract argumentation has become a popular approach for nonmonotonic
inference. The basic idea is to construct a form of aggregates, called
arguments, and to examine how these attack each other (basically, to
construct a directed graph in which the arguments are represented by
vertices and the attack relation is represented by edges). Several
criteria have been stated (called “argumentation semantics”) for
evaluating such a graph, basically to decide which arguments to accept
and which arguments to reject.
Logic Programming is a slightly older approach for nonmonotonic
inference. Here, one starts with a set of rules like “a <- b, c, not d”,
stating that if b and c can be derived and d cannot be derived, then a
can be derived. Again, several criteria have been stated (called “logic
programming semantics”) for evaluating a logic program, basically to
decide which atoms to accept and which atoms to reject.
There turns out to be a striking similarity between these two
approaches. In fact, it is possible to convert a logic program to an
argumentation framework in standard way. Most of the argumentation
semantics can then be shown to coincide with related logic programming
semantics. However, there turn out to be two noticeable exceptions,
questioning whether Abstract Argumentation is indeed a general approach
for nonmonotonic inference.
On the Instantiation of Knowledge Bases in Abstract Argumentation Frameworks
Adam Wyner, University of Aberdeen (joint work with Trevor Bench-Capon, and Paul Dunne, University of Liverpool)
Abstract Argumentation Frameworks (AFs) provide a fruitful basis for exploring issues of defeasible reasoning. Their power largely derives from the abstract nature of the arguments within the framework, where arguments are atomic nodes in an undifferentiated relation of attack. This abstraction conceals different conceptions of argument, and concrete instantiations encounter difficulties as a result of conflating these conceptions. We distinguish three distinct senses of the term. We provide an approach to instantiating AFs in which the nodes are restricted to literals and rules, encoding the underlying theory directly. Arguments, in each of the three senses, then emerge from this framework as distinctive structures of nodes and paths. Our framework retains the theoretical and computational benefits of an abstract AF, while keeping notions distinct which are conflated in other approaches to instantiation.
Interaction with broadcast debates: The Grumpy Old Debate project
Rolando Medellin, University of Dundee
We examine the ways in which software-supported interaction with broadcast debates can be used to reach into social excluded groups of older adults.
Fundamental research on human debate and human-centered computing is combined in this project.
The objective is to understand how older people in particular engage with resources associated with broadcast programmes.
The results of a study where older adults interact with content of a Moral Maze debate are used to define an argumentation protocol.
In the first stage we were able to identify specific debating practices and desirable usability requirements from the perspective of older adults.
Hengfei Li, University of Aberdeen
In this talk, I will introduce probabilistic argumentation
frameworks(PrAFs), which capture uncertainties in Dung abstract
argumentation framework by assigning probabilities to arguments. The
semantics for this framework identifies how likely a set of arguments is
justified under some standard semantics. I will speak about one
shortcoming of PrAFs—the independent assumption—and its restriction.
I will next introduce an extension to PrAFs—probabilistic evidential
argumentation frameworks(PrEAFs)—that allows us to relax the
independent assumption in PrAFs. I will finish this talk by discussing
potential applications of PrAFs and PrEAFs.
Argumentation Theory and Description Logic: an Engineering
Federico Cerutti, University of Aberdeen
Several logics for defeasible argumentation distinguish between “strict”
and “defeasible” rules. Generally, a “strict rule” is a reasoning
pattern that always justifies its conclusion once the premises are
satisfied. The Aristotelian syllogism is an exemplar of this pattern: if
we accept that “Socrates is a man”, and the “strict” rule stating that
“each man is mortal”, then the conclusion “Socrates is mortal” must be
accepted. However, following the seminal work of Reiter (“A Logic for
Default Reasoning”), it has become common to consider rules that allows
exceptions, or, rules that “generally” (instead of “always”) lead to the
stated conclusion. This is the case of the well known rule “generally
birds fly, unless they are abnormal” – penguins are birds, but they
The introduction of such defeasible rules led to what is commonly called
“non-monotonic reasoning”, which aims at overcoming some of the limits
of classical logics. In particular, description logics, a family of
logics for representing terminological knowledge of a domain, are
generally compliant with the axioms of classical logics. Therefore, a
possible way to exploit argumentation techniques in the context of
description logics is to map ontological relationships into strict rules
(e.g. the “ISA” relationships), and other relations in defeasible rules,
in order to deal with inconsistencies. In this talk we would like to
share some preliminary ideas on this topic which seem to support the
choice of using defeasible rules only, instead of mixing them with
strict ones, thus achieving extensibility, history tracking, and
The connection between formal theories of argument and inference on the one hand, and dialogical processes of debate and disagreement on the other is surprisingly understudied. Inference Anchoring Theory (IAT, Budzynska & Reed 2011) provides, for the first time, a well-grounded account of this connection using the theory of speech acts (Searle 1969, 1975). IAT tackles a number of challenging theoretical issues, including argument that uses the statements of others, argument that involves ethos and argument that is established purely in virtue of its dialogical context. IAT is also sufficiently robust to take on the linguistic analysis of natural arguments in unconstrained domains, such as discussion boards on internet fora and moral debates on radio. Finally, IAT is also the lynchpin in extensions to the argument interchange format, a computational standard for the representation of argument by machines, and those extensions are now supporting a raft of innovative, exciting software applications.
Towards the definition of argumentative dialogue types
Mathilde Janier and Patrick Saint-Dizier, IRIT, France
Following the IAT theory elaborated by C. Reed and K. Budzynska,
and in conjunction with these authors, we present an analysis of argumentative
dialogue categories based on a notion of types.
First, basic types are considered and polymorphic types are
introduced to handle the numerous ambiguity problems between categories
between which there is a form of continuum.
Second, given a statement such as ‘Do you believe that P ?’ we show that,
assuming that P is of type assertion (noted as A), it is first modified by a
modal (believe). It still remains of type A: leading to the coupound type: bel(A).
Then, this type may
be coerced into a question (type Q) via the string ‘Do you’, leading to the
compond type Q(bel(A)).
We view these operators as functions operating over types.
If time permits, some linguistic element will be developed to show
how the various elements are realized in language.
Aligning Argumentation Theory with Behaviour Change Mechanisms
Simon Wells, University of Aberdeen
In this short talk I report on preliminary work that aims to effect real change in the context of difficult societal problems. Many such problems stem from the cumulative effects of the individual behaviours of large numbers of people. Digital behaviour change mechanisms are used to support people in forming new habitual behaviours and build on rich psychological models of behaviour dynamics. Argumentation theory has rich models of both argumentation and interaction as well as extensive collections of stereotypical patterns of real-world argumentation. In this work we begin to align elements of pyschological models of behaviour change with models of argumentative interaction. The aim is to increase the motivation of bahaviour change targets, enabling them to make informed and justifiable decisions about their behaviours and to increase the overall effectiveness of behaviour change mechanisms.