Neural networks for legal quantum prediction
University of British Columbia
Master of Laws - LLM
This thesis argues that Artificial Neural Networks (ANN's) have applications within the domain of law and can be built using readily available Artificial Intelligence software for the Personal Computer. In order to demonstrate this, I have built working ANN's using data made available to me by the Faculty of Law Artificial Intelligence Research Project (FLAIR) and Windows™-based ANN software that is commercially available. The reasons for building a system are three-fold. First, a working system is the most graphic demonstration of my above assertion. Secondly, given my background as a practitioner in law, I am concerned to ensure the quick, efficient movement of law-related technology from research laboratory to marketplace. Thirdly, the building of a neural network avails me with the opportunity to analyze comparatively the performance of the ANN's with statistical and expert system models which used the same data and were also built at the FLAIR project. The theoretical foundation for this thesis is the view that, although many legal decisions are often reducible to a set of doctrines, policies or sub-doctrinal rules, certain domains of legal decision-making evade analysis using a rule-based paradigm. Thus, although relational patterns between any given facts and the law exist, they cannot always be described. Therefore in order to build "intelligent" computer programs that can assist the lawyer in his or her work, we should explore the potential of and utilize those tools that can find relational patterns automatically. Having done this, we should attempt to combine the same with those software tools that we understand more fully, namely expert systems and traditional programming methods. The use of hybrid neural network/expert system programs is well developed in many other domains. Legal researchers, however, have yet to even thoroughly examine neural networks as an isolated technology. This thesis is an attempt to right this imbalance.
Law, Peter A. Allard School of