Computer Chess and Legal Knowledge Systems

JURIX '97 - The 10th Conference

Amsterdam 12 December 1997

Reviewed by

Radboud Winkels	and	Robert van Kralingen
University of Amsterdam		Tilburg University
[email protected]		[email protected]

Download

---

This is a Conference Report published on 27 February 1998.

Citation: Winkels R et al, 'Computer Chess and Legal Knowledge Systems: JURIX '97 - The 10th Conference', Conference Report, 1998 (1) The Journal of Information, Law and Technology (JILT). <http://elj.warwick.ac.uk/jilt/confs/98_1jur/>. New citation as at 1/1/04: <http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/1998_1/winkels/>

---

1. Introduction

The Free University of Amsterdam was the venue of the 10th conference of the Dutch Foundation for Legal Knowledge Systems (JURIX) held on December 12, 1997. Of course, an anniversary calls for a special theme. The programme committee chaired by Richard de Mulder selected Computer Chess and AI & Law. More specifically, the question was addressed whether researchers in legal informatics can learn something from the results of computer-chess research. Kasparov's recent defeat by Deep Blue fuelled the discussion as to the ability of computers to intelligently perform complex tasks. The theme was addressed both at a tutorial and at a panel discussion. The other conference lectures were sandwiched between the tutorial and the discussion. In this conference report, the same outline is followed.

2. Tutorials

On the day preceding the conference, two tutorials were held. The first one addressed computer chess, the second one knowledge management. Since computer chess is hardly one of the traditional JURIX themes, some external expertise was invited to give the tutorial. Jos Uiterwijk ( University of Maastricht ) and Wim Pijls ( Erasmus University Rotterdam ) came to the rescue. After an introduction to some basic game-tree search algorithms such as minimax and alpha-beta, an interesting video on the first match between Kasparov and Deep Thought (the name of Deep Blue before the IBM take-over) was shown. The video was followed by a history of computer chess and by a discussion of some recent developments in the area of computers and problem solving. Some games have been 'solved' by number-crunching computers. An example is four-in-a-row. We have obtained 'complete knowledge' on this game. In other games, computers can outperform humans or can at least reach world-champion level. Examples are American checkers and chess. However, in some games, such as Go and Japanese chess, the computer is yet to rise above amateur level. At the moment, computers take a brute-force approach, enhanced with heuristics, to play games. In the future, new techniques such as opponent modelling might improve performance.

The second tutorial (Thanks to Antoinette Muntjewerff for her information on this) addressed a topic receiving increasing attention in the nineties: Knowledge Management. Richard de Mulder (Erasmus University Rotterdam) and Anja Oskamp (Free University, Amsterdam and University of Nijmegen ) were the designated tutors. Knowledge management has been a hype in recent years; many a consultancy firm provides advice and services in this area. De Mulder provided the setting for the tutorial: the lawyer of the future will have to meet new demands. Informatization and globalization will fundamentally alter legal practise and lawyers will have to 'manage' their information and knowledge if they want to be able to provide adequate services. Jurimetrical research can come to the aid of the practising lawyer. To give an example: a client might want his lawyer to predict the outcome of a case. A Conceptual Document Analysis System (CODAS) can be of help in retrieving similar cases that have been classified according to the facts of the case. Such a system provides help at a conceptual level. Oskamp's contribution went along the same lines. She set out by arguing that lawyers use legal information, legal knowledge and legal skills to perform tasks. Next, she addressed legal sources and their electronic counterparts. Basically, we can distinguish electronic legal sources in legal information systems and legal knowledge systems. An important question in this context is whether a modern lawyer can afford not to use information or knowledge systems. In the end, the tutorial turned out to be a lecture, albeit a lecture with opportunities for questions.

3. The Conference

The opening lecture of JURIX'97 was delivered by Ed van Thijn, the former mayor of Amsterdam. Van Thijn stated that in modern society, there is a shift of power from the elected politicians towards the bureaucracy. Since computers have claimed a solid place in the bureaucracy and since the use of computers encourages the rigid, formal application of rules, van Thijn warns against letting computers take over from 'feeling'. Moreover, the shift of power and the use of computers inhibits democratic means of control.

After the rather grim picture painted by Van Thijn, the invited lecture by Abdul Paliwala ( Law Technology Centre , University of Warwick) provided a much more positive approach towards legal informatics. Paliwala made a serious effort in reading the proceedings of the past ten JURIX conferences. In his lecture 'An intellectual celebration: 10 years of JURIX Conferences' he gave an overview of the developments in legal-informatics research. Much has been accomplished in ten years: legal informatics research covers a wide range of topics, many tools and representation formalisms have been developed, etc. With the editors of the JURIX'96 proceedings he saw a shift in the research effort from the development of legal 'expert' systems towards the development of legal knowledge systems aimed at providing support, on the one hand, and fundamental research, on the other hand. As some interesting new directions for legal-informatics research Paliwala mentioned the integration of (representation) techniques, the use of the Internet, and the application of neural nets. However, while doing research, the dangers of a 'jurimetricasation' of society must not be forgotten, i.e. the tendency to specify law in formulas in stead of explicit norms.

The invited lecture was followed by seven presentations of papers that had been accepted to the conference. We will briefly mention the subjects of the papers. The complete proceedings (including Paliwala's lecture) are available (in pdf format) from <http://jurix.bsk.utwente.nl/html/j97.html>. In the first paper session Bart Verheij and Jaap Hage ( University of Maastricht ) presented a paper on an abstract ontology of the law. They look upon the law as a dynamic system in which states of affairs are linked to each other through events and rules. Next in line was Arno Lodder (University of Maastricht). Lodder discussed the distinction between structural and procedural arguments in the law. James Palmer ( University of Oxford ) made the final presentation of the morning session. Palmer argued that the current methods for the representation of legal arguments are not suited to the purpose of generating arguments, nor are they useful in determining which of a set of conflicting viewpoints prevails based on the arguments. In the afternoon the conference continued with a second paper session. In his lecture Pepijn Visser ( University of Liverpool ) discussed the problems occurring when linking heterogeneous systems partly covering the same domain. The problems are caused by so-called mismatches in the systems' underlying ontologies. Visser proposes a classification of ontology mismatches and links these to the problems the mismatches pose. Anneke de Lange and Peter Schipper (Bolesian) presented their experiences in the EC-subsidised Service 2000 project. The project aims at providing better (government) information and services to citizens. John Yearwood ( Ballarat University ) delivered the third afternoon lecture. His research focuses on the possibility of finding legal cases in a database. By making use of the relatively rigid structure of cases, Yearwood attempts to improve search performance. Preliminary results indicate that the approach indeed improves performance. Long-time JURIX participant Trevor Bench-Capon (University of Liverpool) had the honour of giving the final lecture. His contribution dealt with the possibility of using cases in legal reasoning. In his paper Bench-Capon reconstructs the case-based reasoning systems HYPO and combines it with some aspects of its succesor CATO, in an attempt to find out what Case Based Reasoning is all about.

4. Panel Discussion

The day ended with a panel discussion on the conference theme: computer chess and legal knowledge systems, what can legal informatics learn from computer chess? Richard the Mulder chaired the panel. The panel members were Jaap van den Herik (University of Maastricht, University of Leiden and president of JURIX), Abdul Paliwala, Genna Sosonko (a chess grandmaster), and chess journalist Lex Jongsma. The participants introduced the discussion through opening statements. Unfortunately, the focus of the statements and of the ensuing discussion was often on either computer chess or the legal domain. Perhaps the link between chess and law is not as obvious as some (for instance, Jaap van den Herik) might think. Some interesting issues were:

• The introduction of computers has some unintended and possibly unwanted side effects. For instance: chess matches can no longer be suspended, since the players now have the opportunity to have a computer analyse their endgame. In the legal domain, similar unwanted side effects may occur (for instance the jurimetricasation of society mentioned by Paliwala);
• Computer-chess research has shown that machines can be good at some 'intelligent' tasks. However, their approach may be altogether different from a human approach. From this, we can learn that in order for legal informatics to succeed it is not needed to imitate human decision-making processes in order to obtain useful results;
• Computer chess contributes importantly to chess theory. Many errors in, for instance, endgames have been brought to light through application of computers. Perhaps legal informatics can contribute similarly to the development of legal theory.

With the panel discussion, the tenth conference came to an end. However, the anniversary festivities were continued with a reception and a dinner in Amsterdam's dazzling inner city. JURIX will continue in 1998 with the eleventh conference. We look forward to meeting you there.

The proceedings are available from the Web: <http://jurix.bsk.utwente.nl/html/j97.html> or can be ordered from the JURIX secretariat ( [email protected] ).