CLIME: Lessons Learned in Legal Information Serving

This paper presents lessons learned from the CLIME project (1998-2001), aimed at improving the access and understanding of large bodies of legal information through the Internet. Our approach involves explicit representation of the content of the legal sources consisting of three types of knowledge: (1) the domain being regulated (a domain ontology); (2) normative statements about the domain; (3) resolution mechanisms when normative statements conflict (meta knowledge). The “Legal Information Server” (LIS) provides both Conceptual Retrieval (CR) and Normative Assessment (NA) on these knowledge structures, given a user’s input case. We found that, in contrast to ‘traditional’ knowledge systems, cases put to a LIS are often incomplete and not at the right (most specific) level of detail. Therefore we do not only offer a normative judgement based on applicable norms, but also potential exceptions. These exceptions can be (partly) computed off-line based on the representation of the legal sources. Furthermore, we adopted an incremental modelling approach that shows promise. We begin with the domain ontology for CR and extend it afterwards for NA. Experiments showed a strong bootstrapping effect, speeding up modelling. Finally, we conclude that the best moment for the considerable investment of modelling regulations would be during their drafting.