The Problem of Finding a Precedent
One of the traditional problems of legal research has been retrieving prior decisions or cases (precedents) relevant to a current problem. Some of the major computerized legal information services in the world are mainly concerned with providing case law databases.1 Such services rely traditionally on text retrieval systems employing boolean search requests.2
A boolean request can be compared to a hypothesis: The probable properties of the relevant cases are specified, and the system looks for the documents matching these properties. The hypothesis is based on the user's understanding of the problem and may also be seen as a proposal of which properties are common between the current problem, known to the user in great detail, and the relevant cases, unknown to the user. The retrieval is based on a comparison of identity with respect to specified properties between a problem and a case.
In the context of this chapter, it may be important to realize that this comparison of similarity is the basis also for current retrieval systems. In traditional information systems the properties to be specified are the words of the document. Properties of the same category have to be specified for the problem--the search request consists of words or combinations of words, the combinations employing boolean operators like AND or OR, supplemented by a number of other operators like distance operators (adjacent words; words within same sentence or paragraph; words within a certain "distance" from each other, this "distance" also being measured in words; and so on).