Legal expert systems: Words, words words … ?

Abstract

Some special features of the legal domain which affect the development of expert systems are examined, including: why ‘ready made rules’ are a problem, not an answer; the relationship between prescriptive laws and legal source materials; ‘rules’ of interpretation of statutes are not meta-rules; ‘precedents’ of case-law cannot always be treated as rules; the need for access to legal source materials; the relevance of ‘deep’ or causal models to the legal domain; reasons why lawyers must play the main role in developing legal expert systems. DATALEX, a joint research project by the authors which commenced in 1986, is outlined. Its main aims are: (i) to develop a shell which integrates a number of inferencing techniques in a consistent environment, and is itself integrated with a free text retrieval system. The shell, LES (Legal Expert System), includes decision network and nearest neighbour discriminant analysis modules; other inferencing modules are being added; and a free-text retrieval system, AIRS, is being integrated with LES. (ii) to use this shell, and to encourage others to use it, to develop a range of knowledge-based applications in different areas of Australian law. Applications as yet concern intestacy, copyright, the finding of chattels, and other areas of law. (iii) to use these applications as a teaching resource to encourage interest in and understanding of knowledge-based techniques and their potential by Australian lawyers.This 1987 paper was the first paper published about the DataLex Project, which ran from 1985-1995.