I am an academic who is interested in how people reason, how this reasoning can be captured in formal models and how it can be supported and improved using smart technologies. My main areas of investigation are the computational, philosophical and linguistic aspects of argumentation, linking mathematical models with more natural representations of argument and discourse.
In addition to working on argumentation theory, I am also keen to improve argumentation practice by developing tools that can be used to disseminate and analyse complex reasoning involving lots of data. Examples of application areas are legal & forensic reasoning and opinions on the Web.
I am currently working at the Artificial Intelligence Department of the University of Groningen, the Netherlands.
The average criminal case includes a large amount of evidence and hypotheses of which the investigators and decision makers have to make sense. For my dissertation in 2009 (summary), I developed a formal theory that tells judges and police investigators how they can analyze and communicate their reasoning about a case using arguments and stories. This hybrid theory of arguments and stories has been published in book and in journal form.
In a recent short paper, Abductive argumentation with stories, I show how stories and arguments can be further integrated by considering stories as arguments. This integrated theory presents a cleaner and simpler mathematical model than the original hybrid theory, and is fully compatible with well-known argumentation frameworks from AI.
In the October issue of Communications of the ACM, we present an article on the first fully developed implementation of the Argument Web. The Argument Web is a structure of linked argument data underlying the World Wide Web that makes it possible to follow a line of argument (on a particular topic or by a particular person) across disparate forums, comments, editorials, and multimedia resources. In combination with existing and new software tools for argumentation, the aim is to integrate offline and online argument so that it becomes intuitive for a wide variety of users, including mediators, students, academics, broadcasters, and bloggers.
Real-time analysis of arguments in discourse is a cognitively heavy task that, at least at the moment, is nearly impossible to automate. When I was at the University of Dundee's Argumentation Research Group, we used the AnalysisWall (a 6 m2 touch screen) together with bespoke software to analyse the argumentative structure of a 40-minute radio broadcast in real-time. The insights gained from this analysis are used to develop a theoretical as well as a practical approach to automatically generating meaningful argumentative dialogues.
The final analysis of the radio broadcast was directly placed onto the Argument Web and can be viewed here. I also did an individual analysis of the same radio programme earlier, using existing argument mapping tools. This took me about a week. This analysis can be viewed (PDF) by clicking on the picture below.
Postmodernism says that we teach people through stories and narratives rather than by giving them facts and rules. But how exactly do these stories persuade us? This is a question I've been trying to answer together with Trevor Bench-Capon. In a recent short paper, Arguing With Stories, we discuss how stories can be used in Arguments From Analogy, and why common narratives like fables and parables are so convincing. Our aim is to design an implementation that, given a story, automatically generates the possible arguments based on this story.
If you're interested in in the place of stories in AI, I'd recommend to keep an eye out for an upcoming special issue of Literary & Linguistic Computing on Computational Models of Narrative, which I co-edit with Mark Finlayson, Pablo Gervás and Deniz Yuret.