SPECIAL ISSUE ON INFORMATION TECHNOLOGY AND THE LAW II

1 Guest Editorial - IT Applications to the Law

2 The use of object oriented principles to develop intelligent legal reasoning systems G VOSSOS, T DILLON, J ZELEZNIKOW, G TAYLOR

_ ________ 1_1 Legal expert systems: The inadequacy of a rule-based approach J POPPLE

17 The applicability of a knowledge-based system to legal education SD GREGOR, HM RIGNEY, JD SMITH

INFORMATION SYSTEMS

_________ 2_2 Australian involvement in electronic data interchange - 1989 PMC SWATMAN, JE EVERETT

SPECIAL FEATURES

34 Guide to authors --- ------35 Book reviews ---------

Published for Australian Computer Society Incorporated Registered by Australia Post, Publication No. NBG I 124

LegsysinarulappJ. PopDepartmAustrali

The twodescribeobjects odiscussesystemslegal exadopted,knowledsystems

It is aexpert sof lawyecases. Aincorporknowledapproacuseful le

Key Wexpert ssystems.

CR C

Copyright©permission material is given and tissue, and tpermission ManuscriptEarlier versAustralian CInternationa

TR rane. Ityers Th maolveega

es: legabases, folify

e noivaleds). egale thw. cC

l AI examch mons.sefu somehl renal sren

Goeen

l anr do

his ems

EGAal a

al expert tems: The dequacy of a e-based roach

ple ent of Computer Science

an National University

different categories of legal AI system are d, and legal analysis systems are chosen as f study. So-called judgment machines are

d, but it is decided that research in legal AI would be best carried-out in the area of pert systems. A model of legal reasoning is and two different methods of legal ge representation are examined: rule-based and case-based systems. rgued that a rule-based approach to legal ystems is inadequate given the requirements rs and the nature of legal reasoning about new, eclectic approach is proposed, ating both rule-based and case-based ge representation. It is claimed that such an h can form the basis of an effective and gal expert system. ords and Phrases: case-based systems,

ystems, law, legal reasoning, rule-based

1 INThewidlawlaw.thatto s

Lgori

Ldatacasesimphavequwor

Lminof la

(Mlegaan whigatibe uwith

Mdiffetrievdiffeandbetwlegalatte

Tsyst

2 LLeg

ategories: l.2.1, l.2.3, l.2.4, l.2.5, l.2.6, J.l.

1991, Australian Computer Society Inc. General to republish, but not for profit, all or part of this granted, provided that the ACJ's copyright notice is hat reference is made to the publication, to its date of o the fact that reprinting privileges were granted by of the Australian Computer Society Inc. received January 1990; revised October 1990. ions of this paper appear in the proceedings of the Thirteenth

omputer Science Conference (Popple, l 990a) and the l Conference on Computing and Information (Popple, 1990b).

gories: judglike for tlegalsimiprovthoseguiltfrom

2.1. JudThe ide

THE

ODUCTION ge of computer applications in the law is extends from general applications, of use to

, to applications designed specifically for the is paper is concerned only with those systems ke use of artificial intelligence (AI) techniques legal problems. l AI systems can usefully be divided into two cate-egal retrieval systems and legal analysis systems. l retrieval systems allow lawyers to search through s, containing details of statutes and decided r information. AI techniques may be employed to this task (e.g.by searching for keywords which t been input by the user but are deducted to be nt to, or sufficiently related to, the input key-

analysis systems take a set of facts and deter-e ramifications of those facts in a given area

arty ( 1980a) identifies a third category of systems: integrated legal systems. He cites as ple computerized title registration systems ake decisions about people's rights and obli-

It is hard to see why such a system could not lly classified as a legal analysis system, albeit e of the features of a legal retrieval system.)

( 19 59) claims that there is no fundamental ce between these two categories (legal re-ystems and legal analysis systems)-that the

ce is one of degree only. However, Shannon lshani ( 1988) point out that the difference systems based on a "conceptual model of alysis" and text-retrieval systems is that the not "understand" any area of the law.

paper will be concerned with legal analysis .

L ANALYSIS SYSTEMS nalysis systems can be divided into two cate-

ment machines: systems that make a judge-pronouncement (e.g. "X is guilty of offence Y he following reasons ... "); and expert systems: systems that provide advice lar in form to that which a solicitor might ide (e.g. "The facts in this case are similar to in P v D where the defendant was found

y, but the instant case can be distinguished P v D as follows ... ").

gment Machines

a of a judgment machine was raised over

AUSTRALIAN COMPUTER JOURNAL, VOL 23, No. 1, FEBRUARY 1991 11

- CY

thimawa

Agehurelrana vfactreforcoucerwhexasubthe

TchiWe

Tkotm

fmnaaw

tct

MoTiaotIjtgpa

BjWwt

12 TH

rhyevr

ud

oenae

ioa

yeauroor

.2o

LEwoTe

ohhahauroreT

ai

ha sr b

Lsuouisil

ha

Lll

ea

----------- LEGAL EXPERT THE INADEQUA

rty years ago by Mehl (1959). Although such a chine would perform the functions of a judge, it s said that a role for humans remained because: ... the solution to a legal problem may depend upon extra-rational factors, involving the whole of human ex-perience ... (p. 758)

lmost twenty years later, D' Amato ( 1977) sug-sted that a judgment machine could replace a man judge. His proposed machine would take the evant facts as input and produce a number in the ge -1.0 to 1.0 (where a positive number indicates ictory for the plaintiff). Given the multiplicity of tors, he claimed, a result of zero would be ex-mely unlikely. Somewhat begrudgingly, he allowed some vestige of human control. An appeal court ld review all of the machine's determinations in a tain numerical range (e.g. -0.05 to 0.05) within ich the cases would be so close that a re-mination might be required. The review court's sequent decision would then be incorporated into system. he idea of human judges being replaced by ma-

nes has been vehemently criticized. According to izenbaum (1976): he very asking of the question, "What does a judge ... now that we cannot tell a computer?" is a monstrous bscenity. That it has to be put into print at all, even for he purpose of exposing its morbidity, is a sign of the adness of our times. Computers can make judicial decisions ... They can

lip coins in much more sophisticated ways than can the ost patient human being. The point is that they ought

ot to be given such tasks. They may even be able to rrive at "correct" decisions in some cases-but always nd necessarily on bases no human being should be illing to accept . . . . What emerges as the most elementary insight is

hat, since we do not now have any ways of making omputers wise, we ought not now to give computers asks that demand wisdom. (pp. 226-227)

les (1987) agrees: he computer scientists, encouraged by the modern pos-

tivists, fail to recognize ... that law, positive morality nd ethics are inseparably connected parts of a vast rganic whole. Judgments are involved at every stage of he legal process and machines cannot make judgments. n stating that legal rules can be applied without further udgment; that they apply in an all or nothing fashion; hat legal decision making follows the form of the syllo-ism or that it is a pattern-matching routine, the modern ositivists, joined now by the computer scientists take us long a dangerous road. (p. 271)

ut D'Amato sees advantages in replacing human udges by machines:

ould we lose a judge's "judgment," and how important ould such a loss be to our legal system? Surely compu-

Ptslpj

crmmtmsrdpf

2F(sw

mcpttffm

hwtaoa

ucdwc

3Al

ers do not make "judgments" the way humans do, and

E AUSTRALIAN COMPUTER JOURNAL, VOL. 23, No. 1, FEBRUARY 1991

be

A APPROACH ------------

so we would lose the "human" aspect of legal judgments. But what specifically do we lose when we lose the hu-manness of judgments? Is human judgment just a euphe-mism for arbitrariness, discretion, or bias? (p. 1281)

oponents of the idea of automated judges claim at such systems would reduce the cost of the legal stem, find inconsistencies in the law, and provide a el of certainty in the law which does not exist at

esent (as a result of freely-available, automated, icial advisory opinions).

In 1977, D' Amato claimed that his proposal for a mputerized judge was a modest one. Yet, the cur-t state of AI technology is such that no judgment chine has been implemented which can pass judg-nt in any substantial area of law. The ethical ques-n of whether judges ought to be replaced by chines remains a hypothetical one. The author's

mpathies lie with Weizenbaum and Moles. For that son, and because AI technology has not yet pro-ced judgment machines-machines which may ve impossible to build-this paper will, hence-th, be concerned only with legal expert systems.

Legal Expert Systems r the purposes of this paper, a legal expert system S) will be defined as a system that provides an-ers to legal questions which are in a form that one uld expect from a lawyer. his definition excludes AI systems which might

rely be used as tools by a lawyer in coming to legal nclusions or preparing legal argument (e.g. a so-isticated legal retrieval system). This is not to say t a lawyer should not be able to use an LES, merely t the output from an LES should be usable without ther legal analysis. This output should be in such a m that it can be the basis of a lawyer's legal argu-nt in court. he term expert system has become ambiguous. It

s variously been defined as a system that deals th "knowledge" gathered from an expert, a system t can perform at the level expected of an expert, or ystem that can be used by an expert. Applying any, all, of these definitions, an LES—as defined

ove-is an expert system. ESs are not judgment machines; they will not rp judicial power (although their imprudent use ld lead to the relinquishment of some judicial

cretion). The development of sophisticated LESs l not remove the need for lawyers, but it may nge the nature of some legal work.

EGAL REASONING LESs must be capable of legal reasoning, or (at st) of simulating legal reasoning. So all LESs must

based upon a model of legal reasoning.

--

Anthe qor notheornormand dlegal

Theboth that mopen

3.1 OThe tby Ha(1988

Roucom

HmeacatenessissuThe

whichproblMcCaof Mc

Theof corcase represBelowthe mteresttions.levels,(statuor noMcCathe "f

Mcopen-represare oMcCacomplbility

McC'secsumcomtranlega

Theawa

M aseesngld

ildve

owinehewsture t

Ar th pl b l

e i) in multe.

tunt ultea

sttanl, enicace lIfabcainlowIfsirnithsh

Ne men

--------- LEGAL EXPERT THE

y attempt to define legal reasoning soon raises uestion of whether that definition is descriptive rmative. MacCormick (1978) points out that a y of legal reasoning can be both descriptive and ative-that it can be normative in its own right, escribe norms that actually operate within the system. model of legal reasoning adopted in §3.2 is descriptive and normative. Before describing odel, it is necessary to explain the concept of

texture.

pen Texture erm open texture was first used in jurisprudence rt ( 1961 ). According to Shannon and Golshani

): ghly speaking, a concept is open-textured if it defies plete definition. (p. 312) ere we refer to the inherent indeterminacy of the ning of words that are used to describe the predi-s of statutes. Sometimes other factors, such as vague-, may also be considered as part of the open texture e. (p. 312, n. 10) problem of open texture in legislation is one arises in the development of any LES. This em can be demonstrated by reference to rty's TAXMAN project (1980a), and to criticism Carty's approach to open textur.e. TAXMAN project was concerned with the area porate tax law. The basic "facts" of a corporate were captured in a relatively straightforward entation (e.g. a corporation issued securities). this level was an expanded representation of eaning of various entities (e.g. a security in-

) in terms of their component rights and obliga- Above this level-presumably above both although this is not made clear-was the "law" tory rules which classify transactions as taxable n-taxable etc.). Legal analysis, according to rty, is a simple matter of applying the "law" to acts". Carty accepts that some legal concepts are textured, which makes them very difficult to ent. Moles (1987) argues that alI legal concepts pen-textured-that even those concepts which rty claims are easily represented are impossibly ex (in the sense that they are beyond the capa-of machines). Moles complains that: arty appears not to appreciate that 'corporations',

urities', 'property', 'dividends' and so on are not sub-ed 'beneath the law', but are each the products of plex legal analysis. The question of whether certain sactions are taxable or not is intimately tied into that l analysis. (p. 270)

sad thing is that he has not shown the slightest

hiscauposbeibuibuabobeldef

Tallotexabl

3.2Fothewil

A(thanytheuterescas

Atexstaresa mtheinsicagivnifcon

A

4 KThdep

reness of the nature of the legal expertise. (p. 269) ute la

A RULE-BASED APPROACH

oles is opposed to the very idea of an LES, but ttack on McCarty's approach is important be- it emphasizes the problem that open texture for LES design. If all legal concepts are seen as open-textured, as Moles would have it, then ing an LES would be impossibly complex. An LES er must choose an arbitrary level of abstraction which a concept may be open-textured, and which a concept must be considered to be fully d. model of legal reasoning adopted in §3.2 that some concepts in a statute may be open-ed, but assumes that these concepts are amen-o full definition by reference to case law.

Model of Legal Reasoning e purposes of this paper, the following model of

rocess of reasoning with statutes and case law e adopted. awyer examines the facts of the case in question nstant case), and determines which statutes (if apply. These statutes are applied to the facts of stant case. The meaning of a concept in a stat-ay be open-textured, and may determine the

of the application of that statute to the instant

lawyer argues about the meaning of an open-red concept by reference to the facts of the in-case and those of previously-decided cases. The s of some cases are desirable in that they ascribe ning to an open-textured concept which (when atute is applied) leads to a desired result in the t case. No two cases can be completely ident-

given the plethora of facts associated with any case. Some of these differences may be insig-nt, and much of a lawyer's reasoning by analogy rns the legal significance of these differences. awyer argues with cases in the following fashion: the result of a previously-decided case is desir-le, they argue that there are no legally signifi-nt differences between the previous case and the stant case, so the previous case should be fol-

ed. the result of a previously-decided case is unde-able, they argue that there is some legally sig-ficant difference between the previous case and e instant case upon which the previous case ould be distinguished.

OWLEDGE REPRESENTATION ethod of representing legal knowledge in an LES ds upon whether the LES is concerned with stat-

w or case law.

THE AUSTRALIAN COMPUTER JOURNAL, VOL 23, No. 1, FEBRUARY 1991 13

-- A

4.1 SA nuthe ptheseinstaengineffectthe Ba PRKriwa

Shaing thunsat

suutiutetherecthitio

Altcorrecsimilaa statenginresensuggecan binto fwhicherror. scopelem (bto thecomm

4.2 CNo ara newsubjecpreviolaw) iwhichThis merelfollowthe laaccoucases.

Thefrom, statuttexturWhenbuilde14 THE A

to meacofto knca fd n elylshro

hiactfe t

dd

s hyr

DEheaws.

anucthe l

isleas

1 facs

au tonos.

anbativ lfos din

es idee anlt

--------- LEGAL EXPERT SYSTEMS: THE

tatute Law mber of projects have focussed on representing rovisions of a statute as a set of rules. When rules are applied to the facts of a case (i.e. by ntiating previously free variables) an inference e can produce an answer which represents the of the statute on the given facts. For example, ritish Nationality Act has been encoded as

OLOG program (Sergot, Sadri, Kowalski, czek, Hammond and Cory, 1986). nnon and Golshani ( 1988) claim that extract-e rules from a statute in an ad hoe fashion is

isfactory because: bsequent designers/users cannot trace the evol-on of a set of rules from the words of the stat-; rules cannot be mechanically checked for cor-tness; and s ad hoe approach may lead to rule formula-ns which do not work well together. hough it is not possible to check such rules for tness, it must be remembered that lawyers are rly unable to check their own interpretation of ute for correctness. It is up to the knowledge eer to check that the rules are an accurate rep-tation of the statute. Shannon and Golshani st that it is unlikely that a mechanical method e developed for transforming statutory language ormal rules, but methods have been developed (if followed rigorously) reduce the likelihood of The use of such methods would also reduce the for dissimilar rule formulations. The first prob-eing unable to trace the evolution of the statute rules) can be obviated by the sensible use of ents.

ase Law ea of law is covered exclusively by statute. Even statute, which has not specifically been the t of any case, is interpreted in the light of usly-decided cases. Case law (or the common s fundamental to the Australian legal system, relies heavily upon the doctrine of precedent. doctrine states that each decided case is not y an example that later judges may choose to , or to ignore: that case, itself, becomes part of w. This means that any useful LES must take nt of the law embodied in previously-decided problem of representing case law is different and more complex than, that of representing

ory provisions. If a statutory provision is open-ed, the courts give meaning to that provision.

TheuseitiosurGoapp

Tfinina

Thiture

TFINto tof lcasein indFurcase

Itrure

4.2.In rulebecuse"is casean profinilongthe

Aarguencdectherinstresu

faced with this open-textured concept, the LES

r has two options: USTRALIAN COMPUTER JOURNAL, VOL 23, No. 1, FEBRUARY 1991

No

APPROACH

pose the question to the user ("What is the aning of this open-textured concept?"), and to

cept the user's answer as an accurate statement the law; or incorporate, in the knowledge base, expert

owledge as to the meaning that the relevant ses ascribe to the open-textured concept. irst option is satisfactory only if the LES is being by a legal expert who is (presumably) in a pos-to answer the question. This approach would reduce the LES's usefulness. Shannon and ani ( 1988) opt for a combination of both

aches:

s model allows some room for reasoning with the s but relies on the user for input when no clear rence is found. We add depth to our model by filling he basic rules and definitions of the statute with itional factual examples from decided cases. (p. 311)

simplistic solution to the problem of open tex-as severe limitations, as discussed in §4.2.1.

ee, Greenleaf and Mowbray ( 1988), with their R system, take a completely different approach problems posed by the common law. The area which they chose to model is based entirely on They claim that the number of decided cases y given area of law is usually so small that tive tree generation algorithms cannot be used. r, they suggest that it is inappropriate to model

aw using a rule-based system:

not that it is theoretically impossible to write such s, but that it is not the natural way in which lawyers on with cases. (p.232)

Rule-based Systems and Case Law t, it is not possible to formulate production which will adequately represent case law,

se such a rule-based system would be of little a lawyer. It is not just that rule-based reasoning t the natural way in which lawyers reason with

" Such a system may be capable of producing swer (possibly with an attached estimate of its bility). But a lawyer is not interested in a de-e answer-even if it is strongly suggested by a ine of legal authority-because it doesn't assist rmulation of their legal argument. iscussed in §3.2, a lawyer reasons with cases by g that there are no legally significant differ-between the instant case and a previously-d case whose result is desirable, and/or that are legally significant differences between the t case and a previously-decided case whose is not desirable.

amount of reason extraction from an inductive

--- A

rule-baneeds reducewhich an inatexturean appyers remakes

4.2.2 TThe FIfollowiused tgiven which those cby a lwhich these wstatisti

Whebutes wnearesneighbonearestcase ththe casneighbofinds tconclusest othe"and thchanceniques not gre

5 FURTwo dLESs harepreserepreseAlternaeach obutes wbe madinstant

Noneto the system)legal kncombin

SHYSwhich

d sn thept aris

hoagale xtuob oe. pr gomstenikine c

rob is m e

averat isomw se

N paes

oacyeodit oac

Nauose

is na

re a D

ERATorg, H

-------- LEGAL EXPERT THE

se will provide the informaton that a lawyer to argue in this fashion. Hence, attempting to the results of previously-decided cases to rules can be simply added to a statutory rule base is ppropriate approach to the problem of open-d concepts. As Tyree et al. (1988) state, such roach does not reflect the way in which law-ason about cases. But, more importantly, it for an inadequate LES.

he FINDER System NDER system of Tyree et al. ( 1988) takes the ng approach to cases. Expert knowledge is o determine the most important cases in a (fairly small) area of law, and the attributes are of legal importance to the outcome of ases. These attributes are given weights-not egal expert, but by examining the extent to each attribute differs across the cases. Using eighted attributes it is possible to measure

cal nearness (similarity) between the cases. n the facts of the instant case (i.e. those attri-hich are of legal importance) are entered, the t previously-decided case (the nearest ur) is ascertained. If the attributes of the

neighbour are the same as those of the instant en the advice is clear. When the attributes of es differ, FINDER gives details of the nearest ur, and lists the differences. The system also

he nearest case which reached the opposite ion to that of the nearest neighbour (the near-r). That case, and the differences between it

e instant case, are explained. To reduce the of giving bad advice, several statistical tech-are employed to ensure that the nearest case is atly different from the instant oase.

THER RESEARCH istinct forms of knowledge representation in ve been identified. Rules are appropriate for nting statutory law. They can also be used to nt case law, but this approach is inadequate. tively, a set of attributes can be identified for f the relevant cases. By comparing these attri-ith those of the instant case, a statement can e about the common law as it relates to the

case. of the systems discussed in this paper (and, author's knowledge, no previously developed has incorporated both of these methods of owledge representation: a rule-based system ed with a case-based system.

TER (Popple, I 990c) is a prototype of an LES

baseWheconclegal

Th

ItprlearteprinutapItcosysotathpItlethhstItdlaca

6 COThisLES dappra lawmethand appr

ACKThe Drahand N

ThNatioCentSigm

REFD' AM

GeHART

combines a rule-based system with a case- Series

THE

APPROACH -----------

ystem (similar to the case-based FINDER system). e rule-based system encounters an open-textured

, the case-based system is employed to produce a gument as to the meaning of that concept. eclectic approach has a number of benefits:

as the advantage of approximating the ap-ch which a lawyer would take when given a problem. The rules (derived from a statute) applied until the meaning of some (open-red) concept is required. Faced with this lem, a lawyer would turn to the common law rder to further clarify the meaning of the stat-So, too, does SHYSTER: the lawyer's two-stage oach is clearly modeled. es some of the way towards responding to the plaints of those who believe that an expert m can never adequately simulate legal rea-

ng (see Moles' comments in §2.1 and §3.1). By g the search for the meaning of statutes to ommon law, this approach avoids some of the lems inherent in a purely rule-based system. (it would seem) a novel approach to the prob-of knowledge representation in LESs. However, two disparate methods upon which it relies been separately, experimentally demon-ed. not subject to the restrictions on the problem ain which bind previous systems to areas of which are predominantly statute-based, or

-based, but not both.

CLUSION per has discussed previous developments in

ign and has shown how a purely rule-based h is inappropriate if an LES is to be of use to r. A better approach (combining rule-based s with case-based methods) has been outlined is suggested that LESs which incorporate this h will prove to be fruitful objects of research.

OWLEDGEMENTS thor would like to thank Roger Clarke, Peter , Bob Moles, Malcolm Newey, Robin Stanton il Tennant for their comments and suggestions.

research is supported by an Australian l University PhD Scholarship (funded by the for Information Science Research) and by a ata Postgraduate Research Award.

ENCES O, A. ( 1977): Can/Should Computers Replace Judges? ia Law Review, 11(5), pp. 1277–1301. ISSN 0016-8300. . L. A. (1961): The Concept of Law. Clarendon Law

, Oxford University Press (Clarendon), London.

AUSTRALIAN COMPUTER JOURNAL, VOL. 23, No. 1, FEBRUARY 1991 15

---

McCAbastorBruAp

McCAbasNaSta

MacCOClaOx

MEHLchiMeMaProLab

MOLEReaBla

POPPLRulCoMeAusISS

POPPLRulW. ceeformpp.0-9

16 THE A

PLTecencGO

HAity 29(

Nof ISS

ELegStaApHoEdInt19IZFroISB

OGamms ded

piosis

---------- LEGAL EXPERT THE

RTY, L. T. (I 980a): Some Requirements for a Computer-ed Legal Consultant. Technical Report LRP-TR-8, Labora-y for Computer Science Research, Rutgers University, New nswick, New Jersey. pears with minor abbreviations, as (McCarty, 1980b). RTY, L. T. (1980b): Some Requirements for a Computer-ed Legal Consultant. Proceedings of the First Annual tional Conference on Artificial Intelligence (AAAI-80) nford University, pp. 298–300. RMICK, N. (1978): Legal Reasoning and Legal Theory,

rendon Law Series. Oxford University Press (Clarendon), ford. ISBN 0-19-876080-9. , L. (1959): Automation in the Legal World: From the Ma-ne Processing of Legal Information to the "Law Machine". chanisation of Thought Processes, vol. II, pp. 755-787. Her jesty's Stationery Office, London. ceedings of a Symposium held at the National Physical oratory, 24-27 November 1958.

S, R. N. (1987): Definition and Rule in Legal Theory: A ssessment of H. L. A. Hart and the Positivist Tradition. Basil ckwell, Oxford. ISBN 0-631-15342-X. E, J. ( 1990a): Legal Expert Systems: The Inadequacy of a e-based Approach. Proceedings of the Thirteenth Australian mputer Science Conference (ACSC-13), Monash University, lbourne, 7-9 February, pp. 303-313. tralian Computer Science Communications, vol. 12, no. 1. N 0157-3055. E, J. (1990b): Legal Expert Systems: The Inadequacy of a e-based Approach. In Aki, S. G., Fiala, F., and Koczkodaj, W. (editors), Advances in Computing and Information: Pro-dings of the International Conference on Computing and In-

ation (ICCI-90), Niagara Falls, Canada, 23-26 May, 348-351. Canadian Scholars' Press, Toronto. ISBN

POP

SER

SHA

TYR

WE

BIJ

Cohaestcomcatthe

21627-70-X.

USTRALIAN COMPUTER JOURNAL, VOL. 23, No. 1, FEBRUARY 1991

exper

A RULE-BASED APPROACH -------------

E, J. (l 990c): SHYSTER: A Hybrid Legal Expert System. hnical Report TR-CS-90-06, Department of Computer Sci-e, Australian National University. T, M. J. SADRI, F., KOWALSKI, R. A., KRIWACZEK, F.,

MMOND, P. and CORY, H.T. (1986): The British National-Act as a Logic Program. Communications of the ACM,

5), pp. 370-386. ISSN 0001-0782. NON, D. T. and GOLSHANI, F. (1988): On the Automation Legal Reasoning. Jurimetrics Journal, 28(3), pp. 305-315. N 0022-6793. E, A. L., GREENLEAF, G. and MOWBRAY, A. (1988): al Reasoning: the Problem of Precedent. In Gero, J. S. and nton, R. (editors), Artificial Intelligence Developments and

plications, eh. 16, pp. 231-247. Elsevier Science (North-lland), Amsterdam. ited selection of papers to the Australian Joint Artificial elligence Conference, Sydney, Australia, 2-4 November 87. ISBN 0-444-70465-5. ENBAUM, J. (1976): Computer Power and Human Reason: m Judgment to Calculation. W. H. Freeman, San Francisco. N 0-7167-0464-1.

RAPHICAL NOTE es Popple is a PhD student at the Department of

puter Science, Australian National University. He egrees in law and computer science, and is inter- in the areas where those two disciplines overlap: uter applications in the law, and the legal impli-ns of computer technology. Mr Popple's doctoral research focusses on the development of legal

t systems.