Thomas F. Gordon

The Zeno Argumentation Framework: argument mediation on the World-Wide Web

The Zeno Argumentation Framework is a formal model of argumentation based on and generalizing the models of Pollock, Rittel, and Toulmin. The basic elements of the model are issues, positions, pro and con arguments and preferences. Preferences are a special kind of position, and may be argued about in the same way. Using these elements, the concept of a dialectical graph is defined, for modelling the state of a debate after each move. An evaluation function on the positions in a dialectical graph, using a set of evaluation (or "proof") standards for issues of the graph, is formally defined. The evaluation function determines the current acceptability of each position and may be viewed as another formal specification of the notion of argument "validity". The evaluation function is "nonmonotonic" in the sense that extensions of a dialectical graph can cause some positions which had been acceptable to become unacceptable, and vice versa.

Zeno was designed to be used in computer systems for supporting planning among multiple participants with conflicting views and interests. We call such systems "mediation systems". In the European GeoMed project, we are building a discussion forum for the World Wide Web, based on the Zeno Argumentation Model, which is to be tested in pilot projects in the Domain of city and regional planning.

A software engineering perspective of the Zeno Argumentation Framework is presented, using the Object Modelling Technique (OMT). This methodology provides diagramming methods which are suitable for illustrating the data flow and dynamic aspects of dialectical processes, which are inherently procedural.

Unlike most accounts, the Zeno Argumentation Framework does not extend or adapt an existing formalization of propositional logic. We argue that the argument structure of a natural language text is at a different level of granularity and abstraction than its logical structure. There is no need to reduce one to the other. They are complementary; each has its own range of applications and utility. But neither is inherently more foundational, or more formal.

In the discussion, a plea will be made for a shift in emphasis of argumentation theory from the post facto analysis and evaluation of natural language texts towards providing better theoretical support for mediating argumentation processes in real time, especially for applications in the context of computer mediated communication.

Maastricht Argument Day: abstracts