Dialectical Argumentation Machines

Humans use argument to express disagreement, to reach consensus and to both formulate and convey reasoning. The theory of argument has found wide application in artificial intelligence, providing mathematical structures for automated reasoning, communication protocols for distributed processing and linguistic models for natural language processing. A key stumbling block, however, has been joining together models that focus on abstract, mathematical relationships with those that focus on concrete, linguistic relationships. The first objective of this project is to develop for the first time a theoretical account that connects static, “monologic” argument with dynamic, multi-person, “dialogic” argument and ties together abstract, mathematical models with concrete, linguistic representations.

Furthermore, models of argument have been predominantly confined to the lab. Our goal is to translate the research advances into high profile, large scale deployments using partners with enormous user bases. Prototype systems in this area have been sufficient to demonstrate the unique advantages of practical argumentation systems to potential users of this research such as those within the broadcasting domain. There is a demonstrated public demand for argument-based exploration of current issues with complex scientific and ethical dimensions, demonstrated, for example, by the longevity and success of high profile programming featuring topical issues discussed in a stylised argumentative debate format. The second objective of this project is to develop the theory into implemented components that can form a foundation for application development to support actual programmes with prototype testing Unique advantages afforded by the technology will allow users to interact with the programme material as if they were themselves contributors, allowing arguments to be probed, tested and extended, and the distinction between in-programme and post-programme content to be blurred. The interaction metaphor shifts from ‘message-then-next-message’ to ‘question-answer-riposte-challenge…’.

The rich structure is natural for users, and provides rich metadata for programme-makers. Finally, in 2007 an exciting vision of the “world-wide argumentation web” (WWAW) was laid out, in which systems such as those constructed to work alongside practical prototypes could interact, both with each other and with other debate and argumentation systems, both populist and academic. Argument fragments, expressed as resources on the Semantic Web, can cross-refer, allowing different debating systems to navigate the WWAW according to various rules of dialogue captured by dialectical games. To bring this vision of the WWAW into reality, the third and final objective of the project is to allow execution of arbitrary dialogue games on a platform that provides interfaces for human players, and both interfaces and control for computer players of dialogue games.

Our application areas represent the diversity of domains in which argumentation plays a pivotal role. We want to harness the enormous channel to market and the high-profile reference case that is offered by collaboration within mass media, in the context of mediation, with applications in education, and with the judiciary.

More details will appear here as the project gets underway, but in the meantime, you can read the abstract on the EPSRC website. This project has brought the opportunity to increase the size of the group with the addition of three new members, Mark Snaith, who was one of my honours project students last year working on OVA and worked with ARG:dundee over the summer developing the newlook argdb and integrating it with OVA, John Lawrence, who originally developed MAgtALO whilst an MSc student working for Chris, and Floris Bex who recently completed his Ph.D thesis, entitled “Evidence for a Good Story” supervised by Henry Prakken.