FOA Home | UP: Inference beyond the \Index

FOA(The Law)

The careful reader will have noticed that there have been an unusually large number of examples drawn from the legal domain. There are a number of things it can teach us about general principals of FOA [REF398] . The Common Law tradition is very old, and has been connected to an organized corpus of documents in free text for a very long time. As the Doonesbury cartoon (part of Gary Trudeau's Bicentennial series) in Figure (figure) suggests, legal documents help to demonstrate just how long prose can live beyond its drafting. Hafner was one of the first to recognize this, manually representing a wide range of attributes for a small number of documents in the legal domain [REF131] . This work has now become a part of a larger effort within AI to model the legal reasoning process (e.g., reasoning by analogy [Ashley90] ). Here we are most concerned with what can we learn from lawyers who have FOA the Law which might generalizes to other corpora and searchers.

First, the fact that judicial opinions have been written for so long and in such a particular ``voice'' has meant that it is also possibel to consider special linguistic characteristics of this legal genre [Goodrich87] [Levi82] .

Second, notions of citation are especially well-used within this corpus. Simple concepts like impact have been described [REF1074] , and theories of legal citation proposed [REF128] [REF1069] [REF1071] . Obviously, manipulation of access to the legal printed record (for example by controlling which judges' opinions are made available!) has enormous political ramifications [Brenner92] . This become even more true as recent consolidation of the media industry means that one or two corporations effectively control the entire process of legal publication.

Third, the backbone of the legal process is an adversarial argument. This dialectic is often explicit, for example as marked by the { cf., but cf.} citation conventions (cf. Section §6.1 ). The presense of such syntactic marker makes it conceivable to analyze polazrization across an entire legal literature. Of course the arguments contained in briefs and opinions contain a great deal more structure than simple opposition. The analysis of legal ARGUMENT STRUCTURES , in conjunction with the textual foundations of Common Law, is perhaps the most important feature of the legal domain to FOA. One the one hand, it is possible to model individual documents and their {\em logical} features so as to reason about them [REF131] [REF124] . Special DEONTIC LOGICS have been developed especially to deal with the concepts of ``rights'' and ``obligations'' that are at the heart of many legal relations [REF730] .

{Statistical} analyses of large document corpora may seem contrary to {\em logical} analyses of the arguments contained in each of them, and in fact this chasm runs very deep. Not only does it suggest quite different technology bases from the arsenal of (roughly inductive vs. deductive) AI techniques, but reflects a tension within the law itself. Rose [REF1113] refers to a spectrum of legal philosophies ranging from ``formalism'' to ``realism.'' On the one hand, many legal documents certainly seem to function logically, with careful definitions and reasoning that Langdell [Langdell87] has idealized as ``mechanical jurisprudence.'' At the same time analyses such as the Critical Legal Studies [Unger83] have helped to demonstrate that the Law is just another social process.

It is exactly this dual nature of the Law and hence legal texts that make it especially interesting as an example of FOA [Nerhot91] . Individual applications include LITIGATION SUPPORT systems which allow lawyers to search through the truckloads of documents involved in extended trials. On a much larger scale, systems like West Group's WestLaw and Reed Elsevier's LEXIS systems provide access to the bulk of statuatory and case law to all practicing lawyers [REF1064] [REF733] .

Top of Page | UP: Inference beyond the \Index | ,FOA Home

FOA © R. K. Belew - 00-09-21