Information Technology, the Law and Fuzzy Logic

• Autore: Mario G. Losano
• Pagine: 7-12

Page 7

This Issue of Informatica e diritto presents the papers submitted to the conference entitled ´The Computer and Vagueness: Fuzzy Logic and Neural Netsª, held in Munich on 20 November 1992. Organized by the Italian Cultural Institute of Munich, together with the Institute of Philosophy of Law and Legal Informatics of the University of Munich and with my Chair of General Theory of Law at the University of Milan, this conference marked the culminating point of a successful process of international cooperation, begun in 1988, The original impetus behind this process had been provided by Umberto Rinaldi the then director of the Italian Cultural Institute and by Giuseppina Lorenz-Perfetti, of the same Institute.

The meetings held over this five year period enabled us to keep abreast of the latest developments in legal informatics and computer law. Unfortunately, recent financial cuts have curtailed - at least for the time being - this valuable custom. Looking back at our four Munich meetings, it is now clear that the conference on fuzzy logic originated in a context in which so much importance was attached to being up-to-date that often we were even ahead of our time.

The first conference, in November 1988, was devoted to the fledgling CD-ROM industry and compared the state of the industry in Germany and Italy, Speakers focused not only on matters relating to the production of compact disks, but also on the technologically advanced use of optical supports so that the use of CDs could be extended beyond that of providing mere storage for materials already available in print. In this connection. Lothar Philipps examined the feasibility of ´retrieval enhancement and Andreas W. Vichr described how hypertext systems could also be used en this kind of support. The eight Italian and German papers have since been published in Italian¹.

Page 8

The conference on neural nets, held on 16 and 17 March 1990², arose from dissatisfaction with the way In which traditional programming techniques were being applied to the law. For even though law remains the human science that has tried the hardest to operate with precisely defined words and concepts. It has not been able to free Itself completely from the vagueness that characterises not only human life but also the legal relationships that sustain it. Rules and regulations leave a margin for discretion to those whose job It Is to enforce them: and It Is just this grey area of discretion that can often constitute an Insurmountable obstacle to the use of the computer. We can get around this In two ways: either by simplifying rules and regulations to the level of non-discretionary structures (here we are assisted by law-drafting techniques, to which No. 1/1993 of this journal was devoted), or by employing types of programming that can cope with vagueness In rules and regulations and the resulting discretion necessary In decision making. The classical example here Is reasoning by analogy³.

Neural nets differ from traditional computer technology In that they do not Involve microprocessors (or, rather, they only Involve a few, carefully co-ordinated microprocessors). Instead, their operation Is based on the feedback ratio of a number of ´neuronsª. These microprocessors do not have to be programmed but are trained using specific cases on which they then model their subsequent decisions.

Neural nets have proved good at pattern recognition, especially when what Is required Is a decision based on Incomplete or Indecisive" data: theyPage 9 proceed by constructing or reconstructing the case in question by drawing upon the model already known to them.

So far neural nets have mainly been used in the field of technology, for example, for diagnosing engine noises. However, the idea of bringing their strenghts to bear on legal models was a perfectly natural development, Indeed, neural nets perform a task that is also undertaken by case law (assuming a narrow definition, entailing the work of the law courts): they reach a decision based on degrees of similarity, relying on the combination of a variety of points - of view that in themselves...