On the Issue of Argumentation and Informedness

Formal argumentation theory has been applied to study a whole range of different questions, like “what to accept” [9], “how to come to a joint position” [8] and “how much do different positions differ” [3]. In the current document, we will study a similar question, which can be summarized as “who knows more”. In the context of nonmonotonic reasoning, determining who knows more is far from trivial. One cannot simply compare the sets of entailed conclusions. After all, the fact that the set of entailed conclusions of one agent is bigger (superset) than the set of entailed conclusions of the other agent might be due to the fact that the latter agent has information that invalidates one of the inferences made by the first agent. In this case, it would not seem reasonable to claim that the former agent has more knowledge (or is better informed) than the second agent. Another approach would be to compare not the sets of entailed conclusions, but instead to compare the contents of the knowledge bases. This, however, leads to problems like what to do when two knowledge bases are syntactically different but semantically equivalent. Furthermore, it could very well be that an agent has information that another agent knows to be inapplicable (say, reading a newspaper that another agent knows to be unreliable). Therefore, measuring the raw contents of the knowledge bases is not necessarily appropriate to determine which agent knows more. Yet, the issue of coming up with a suitable criterion for determining who knows more is an important one. One of the reasons that that in a market in which the product that is traded is information and analysis (as is assumed in the LAAMI project) agents can have an incentive to pretend to be more knowledgeable than they actually are, in order to be able to charge a higher fee for their services. However, in order to study questions like whether the agents who are actually more knowledgeable are also economically more successful, we first need to have a formal criterion for determining who the most knowledgeable agents actually are. In the literature on formal logic, issues of reasoning about knowledge (epistemic reasoning) have traditionally been studied using modal logic, in particular by applying modalities that satisfy the KD45 or KT45 axiomatizations. The