Technical Report UU-CS-2007-048, Department of Information and Computing Sciences, Utrecht University, December 5, 2007 (revised September 15, 2008).
An extended abstract of this report has been presented at the JURIX-07 Workshop on Modelling legal Cases, Leiden (The Netherlands), December 12, 2007. A journal version of this extended abstract has appeared in Artificial Intelligence and Law, December 2008 (special issue devoted to the JURIX-07 workshop edited by Katie Atkinson).
This report presents a formal reconstruction of a Dutch civil legal case in Prakken's formal model of adjudication dialogues as instantiated with a rule-based logic for defeasible argumentation. The object of formalisation is the argumentative speech acts exchanged during the dispute by the adversaries and the judge. The goal of this formalisation is to test whether AI & law models of legal dialogues in general, and Prakken's model in particular, are suitable for modelling particular legal procedures; to learn about actual legal procedures; and to learn about the process of formalising an actual legal dispute.
I have visualised the reconstruction of the case with a number of files produced with Araucaria, a programme designed by Chris Reed and Glenn Rowe. These files only visualise the arguments put forward by the parties and their dialectical relations; Figures (1-3) in the report also visualise the speech act structure of the dispute.
At this page both the .aml files, to be used within Araucaria, and the .jpg files produced with Araucaria from the .aml files, can be downloaded for inspection. Needless to say that all rights are reserved.
In the Araucaria charts some compromises had to be made with respect to the report's formal reconstruction of the case, since Araucaria does not support explicit visualisations of undercutting, inapplicability and priority attacks. These attacks have been visualised as attacks on the conclusion of the attacked argument. The Araucaria charts also contain an element not made explicit in the report's reconstruction of the case, viz. a visualisation of the names of the argument schemes implicitly used in the arguments. Within Araucaria I have defined and visualised five such schemes, namely, statutory rule application, counts-as rule application (for classification and interpretation arguments), default rule application (for applications of empirical generalisations), and the above-mentioned schemes from witness testimony and temporal persistence.
In the initial phase of the dispute, plaintiff and defendant present and discuss their legal claims. The files claimphase.aml and claimphase.jpg (corresponding to Figure 4 in the report) show plaintiff's main argument in this initial phase. The files claimphase2.aml and claimphase2.jpg (corresponding to Figure 5 in the report) join it with defendant's main counterargument in the same phase. Plaintiff's secondary attacks on defendant's main counterargument (Figure 6 in the report) are visualised in the files claimphase4.aml and claimphase4.jpg. Conversely, defendant's secondary attacks on plaintiff's main argument (Figure 7 in the report) are visualised in the files claimphaseattack1.aml and claimphaseattack1.jpg. All these arguments are combined in claimphase3.aml and claimphase3.jpg (corresponding to Figure 8 in the report).
The next two argument graphs visualise the `evidential' discussion, that is, the discussion about matters of fact. The files decisionmainarg.aml and decisionmainarg.jpg (Figure 9 in the report) visualise the judge's deciding argument in favour of defendant, while decisionmainargattack.aml and decisionmainargattack.jpg (Figure 10 in the report) extend it with the judge's `internal dialectic', i.e., the counterarguments he has himself considered, as well as their refutations. It should be noted that for reasons of space these charts and the one below do not repeat arguments against a certain witness as attacks on all statements made by that witness.
Finally, a full overview of all arguments and their dialectical relations from the evidential discussion is given in decisionfull.aml. This file nor its .jpg version are included in the technical report since they are too detailed to be readable on A4 paper. For similar reasons I have not even attempted to make a single .aml file that combines the legal and evidential phases of the dispute.