University campus Windmill

SIKS/JURIX Masterclass on Normatics

10 December, 2003

On Wednesday December 10, a masterclass on "Normatics" will be at Utrecht University, jointly organised by JURIX and SIKS. The speakers will be Marek Sergot on normative institutions, Trevor Bench Capon on models of legal argumentation, Giovanni Sartor on law and intelligent agents, and Tom van Engers on large-scale applications of legal knowledge-based systems. Attendance (including lunch) is free for SIKS-Ph.D.-students and for SIKS and JURIX members, while other attendees will be asked to pay a fee of 30 EURO.

Location

Faculty of Law, CIM, Kromme Nieuwegracht 80 (entrance on Herenstraat 24), Room 0.06. How to reach CIM printable map

Registration

  • For SIKS and JURIX members: please register before December 5th, 2003 by email with jurix03@jurix.nl or by telephone with Henry Prakken, tel. +31-30-2532313.
  • For other attendants: see the JURIX-03 conference registration page.

Programme

9.30-11.00: Marek Sergot (Department of Computing, Imperial College London): Action Languages for Modelling Norms and Institutions. abstract presentation
11.00-11.30: Coffee break
11.30-13.00: Trevor Bench-Capon (Department of Computer Science, The University of Liverpool): Models of legal argumentation. abstract presentation
13.00-14.00: Lunch
14.00-15.30: Tom van Engers (Faculty of Law, University of Amsterdam / Dutch Tax and Customs Office): Normative Systems Design, Dealing with Complexity: A Design Methodology for Large-scale Applications of Legal Knowledge-based Systems. abstract presentation (part 1) presentation (part 2)
15.30-16.00: Tea break
16.00-17.30: Giovanni Sartor (Faculty of Law, University of Bologna): Cognitive Automata and the Law. abstract presentation (part 1) presentation (part 2) paper

Abstracts

Marek Sergot (Department of Computing, Imperial College London): Action Languages for Modelling Norms and Institutions.

We want to be able to represent and reason about procedures and protocols in institutions---what actions are possible, permitted, valid at each stage, what rights, powers, obligations are created and destroyed as the protocol/procedure progresses, what kind of normative or institutional relations are created when the protocol/procedure reaches completion. We want to be able to determine from a recorded narrative of what actions have been performed what actions are now possible, permitted, valid, and what their effects will be; and we should also like to be able to prove general properties of these protocols (termination, `fairness', that such and such a protocol has a particular property, that two protocols have identical effects, and so on). Typical examples include auction protocols, rules of procedure, argumentation/dialogue games, negotiation protocols, contract formation, business rules and procedures, procedural law.

The talk will present the development of an extended form of the language C/C+ (Giunchiglia, Lee, Lifschitz, McCain, Turner), which is a formalism developed in AI for specifying and reasoning about the effects of actions and the persistence (`inertia') of facts over time. The aims of the presentation are (i) to identify the basic concepts required for modelling norm-governed institutions, (ii) to present a practical action-based formalism and its associated computational tools, and (iii) to compare briefly with other approaches in the literature.

Trevor Bench-Capon (Department of Computer Science, The University of Liverpool): Models of legal argumentation.

In this talk I will begin by discussing some distinctions between the notions of proof and argument. I shall then look at some approaches to argument based on cases in AI and Law, starting with the HYPO system of Rissland and Ashley and the CATO system of Ashley and Aleven, and going on to look at rule based approaches to the problem. I shall then look at models of argument form, illustrated by systems using Toulmin's argument schema, and models of systems of arguments based on Dung's framework. I hope to conclude by briefly describing my recent work extending Dung's framework to treat arguments based on different social purposes.

Tom van Engers (Faculty of Law, University of Amsterdam / Dutch Tax and Customs Office): Normative Systems Design, Dealing with Complexity: A Design Methodology for Large-scale Applications of Legal Knowledge-based Systems.

Designing and implementing normative systems within governments and their public administrations is a complex issue. The complexity is first of all due to the intrinsic complexity of the legal basis for governmental actions and this consequently reflects in the governmental processes and supporting systems. Another causing factor is the overwhelming number of stakeholders that are involved when new or changed legislation has to be implemented. I will present an approach that was inspired by knowledge engineering that supports the design and implementation of complex normative systems. Important practical issues adressed are: improved legal quality, reduction of time to market and reduction of total cost of ownership.

Giovanni Sartor (Faculty of Law, University of Bologna): Cognitive Automata and the Law.

Our technological society is populated by more and more complex artificial entities, which exhibit a flexible and multiform behaviour. Such entities increasingly participate in legally relevant activities, and in particular in negotiation. Today many contracts are made by computer systems, without any human review. In particular, this happens through software agents (SAs), who may execute autonomously the mandate that has been assigned them, without any subsequent contact with their human users.

In this context, we naturally tend to apply also to artificial entities, and especially to SAs, those interpretative models which we apply to humans. In particular, we tend to explain the behaviour of such entities by attributing them mental states (beliefs, desires, intentions ...). Consequently, we tend to qualify their actions through legal notions which presuppose the attribution of mental states. Consider, for example, the possibility that a computer system enters into a contract (wants to make the contract, and to realise the legal results it states), is the object of a fraud (is cheated), makes a mistake (has a false belief), harms somebody with malice (intentionally), etc.

Our natural tendency to attribute mental states to artificial systems, and to apply the consequent legal qualifications, can be contrasted according to the view that all mentalistic concepts only apply to humans. This thesis would imply the necessity to undertake an extensive review of the existing legal notions, in order to apply them also to the relation with or between artificial systems. We would need to eliminate from such concepts any connection with mental or spiritual attitudes: the element of will needs to be eliminated from the notion of a contract, the element of having a false belief from the notion of a mistake, the element of producing such false belief from the idea of misrepresentation, the element of intention from the notion of malice ...

We will affirm that this choice does not exhaust the available options. It is possible to interpret mental concepts in a flexible and neutral way, so that they become applicable also to some types of artificial entities. This would allow us to preserve both the spirituality and the unity of the law, even in a society which is increasingly characterised by automated information processing.