PADUA: a protocol for argumentation dialogue using association rules

Abstract

Abstract:We describe PADUA, a protocol designed to support two agents debating a classification by offering arguments based on association rules mined from individual datasets. We motivate the style of argumentation supported by PADUA, and describe the protocol. We discuss the strategies and tactics that can be employed by agents participating in a PADUA dialogue. PADUA is applied to a typical problem in the classification of routine claims for a hypothetical welfare benefit. We particularly address the problems that arise from the extensive number of misclassified examples typically found in such domains, where the high error rate is a widely recognised problem. We give examples of the use of PADUA in this domain, and explore in particular the effect of intermediate predicates. We have also done a large scale evaluation designed to test the effectiveness of using PADUA to detect misclassified examples, and to provide a comparison with other classification systems.

A history of AI and Law in 50 papers: 25 years of the international conference on AI and Law

“…This is a gratifyingly high, but rather similar to other techniques developed specifically in an AI and law context, such as neural nets ( [28] and see 5.6: disappointingly no neural net comparison was used in [72]), the argumentation based theory construction of Alison Chorley [78] and the argument mining approach of Maya Wardeh [278]. Some reasons why cases are misclassified are given; that a case may have features which make the decision sui generis, that the decision is simply wrong (fortunately rare) that the set of factors do not capture all aspects of the case, or that the ascription of a factor may be a matter of debate.…”

“…At the same time, Bench-Capon (and the field) remains interested in the learning of rules underlying legal decisions, a recent contribution being the PADUA system [278]. In this work, the same legal conditions are used as in the 1993 paper, this time using training sets that can contain errors.…”

On a Computational Argumentation Framework for Agent Societies

“…This work pools the opinions of several agents that have access to different datasets to predict the classification of a new example. In a subsequent research, the PADUA protocol has been extended to allow multiagent dialogues by proposing the PISA protocol [50,51]. Here, the authors tackle issues like the dynamic creation of a group, the selection of a group leader and intra-group consultation to suggest moves.…”

Agents that Learn How to Generate Arguments from Other Agents

“…Moreover, the rules obtained are only useful in the context in which they were discovered. Similarly, in [30] and [29] arguments are pooled from the agent's experience by means Figure 2 The relationships between number of arguments observed, number of facts in the context and the execution-time.…”