Title : Legal Expert Systems: A Humanistic Critique of : Mechanical Legal Inference Author : Andrew Greinke Organisation : The Australian National University Keywords : Computers and law; expert systems; artificial : intelligence; computerised decision-support systems Abstract : The author surveys a wide range of computerised expert systems and shows that they invariably rely on pattern-matching and rule application strategies which have been embodied in their inference mechanisms and knowledge representations. This computational approach is argued to be unsuitable for use with law which presents a domain of intractable complexity arising out of the need to refer to social context and human purpose in resolving legal issues. The author concludes that a better use for computation in legal applications is in the form of decision- support systems that leave legal inference to human agents. Contact Name : The Editors, E Law Contact Address: Murdoch University Law School, PO Box 1014, : Canning Vale, Western Australia, 6155 Contact Phone : +61 09 360 2976 Contact Email : elaw-editors@csuvax1.murdoch.edu.au Last Verified : Last Updated : Creation Date : 29 November 1994 Filename : greinke.txt File Size : 126KB File Type : Document File Format : ASCII ISSN : 1321-8247 Publication Status: Final COPYRIGHT POLICY: Material appearing in E Law is accepted on the basis that the material is the original, uncopied work of the author or authors. Authors agree to indemnify E Law for all damages, fines and costs associated with a finding of copyright infringement by the author or by E Law in disseminating the author's material. In almost all cases material appearing in E Law will attract copyright protection under the Australian _Copyright Act 1968_ and the laws of countries which are member states of the _Berne Convention_, _Universal Copyright Convention_ or have bi-lateral copyright agreements with Australia. Ownership of such copyright will vest by operation of law in the authors and/or E Law. E Law and its authors grant a license to those accessing E Law to call up copyright materials onto their screens and to print out a single copy for their own personal non-commercial use subject to proper attribution of E Law and/or the authors. EMAIL RETRIEVAL: message "get elaw-j greinke.txt" URL gopher://infolib.murdoch.edu.au:70/00/.ftp/pub/subj/law/jnl/elaw/refereed/ greinke.txt ftp://infolib.murdoch.edu.au/pub/subj/law/jnl/elaw/refereed/greinke.txt -------------------------------------------------------------- LEGAL EXPERT SYSTEMS: A HUMANISTIC CRITIQUE OF MECHANICAL LEGAL INFERENCE by ANDREW GREINKE BComm (Hons) LLB (Hons) CPA Lecturer, Department of Commerce The Australian National University "Suppose I am in a closed room and that people are passing in to me a series of cards written in Chinese, a language of which I have no knowledge; but I do possess rules for correlating one set of squiggles with another set of squiggles so that when I pass the appropriate card back out of the room it will look to a Chinese observer as if I am a genuine user of the Chinese language. But I am not; I simply do not understand Chinese; those squiggles remain just squiggles to me." [*] 1: INTRODUCTION Computerisation of the legal office is an ongoing process. The range of non-legal applications now in common use include word processing, accounting, time costing, communication and administration systems. [1] More recently it has been demonstrated that computers can be used as research tools, particularly in the retrieval of primary legal materials. Prominent examples include the LEXIS, SCALE and INFO1 databases, now familiar to many practitioners. [2] Some moves have also been made towards "conceptual" text retrieval systems. [3] Flushed with successes in projects such as DENDRAL, [4] PROSPECTOR, [5], and MYCIN, [6] computer scientists have now turned to law in order that they might "widen their range of conquests". [7] The interaction between computers and the law has now spawned a large and disparate discipline, boasting eight centres for Law and Informatics in Europe, as well as growing numbers of similar centres in North America, Japan and Australia. The "resource" of expert legal knowledge, "often transitory, even volatile in nature" is seen worthy of nurture and preservation. The use of legal expert systems is seen capable of preserving indefinitely and placing at the disposal of others the wealth of legal knowledge and expertise. [8] The idea is not new, being anticipated by writers such as Loevinger [9] and Mehl [10] as early as 1949. Yet lawyers have generally greeted "legal expert systems" - seen by some as the natural progression in the use of computers - with apathy, ignorance or resistance. [11] This article argues that such opposition is justified when proper regard is had to the implications arising from the computational foundation for such systems. It is necessary for the following analysis to clearly distinguish two fundamentally distinct classes of computer applications to law: decision support systems, and expert systems. [12] "Decision support systems" are powerful research tools or "intelligent assistants" designed to support decisions taken and advice given by human experts. "Legal expert systems" are designed to make decisions and provide advice as would a human expert. Richard Susskind, a British researcher whose work [13] constitutes the major theoretical grounding of legal expert systems, states: "Expert systems are computer programs that have been constructed (with the assistance of human experts) in such a way that they are capable of functioning at the standard of (and sometimes even at a higher standard than) human experts in given fields . . . that embody a depth and richness of knowledge that permit them to perform at the level of an expert."[14] Legal expert systems are a type of knowledge based technology. With the explosion of applications, "expert system" is quickly becoming an imprecise term. [15] The definition used by Feigenbaum will be acceptable for the type of systems examined in this article: "An intelligent computer program that uses knowledge and inference procedures to solve problems that are difficult enough to require significant human expertise for their solution. Knowledge necessary to perform at such a level, plus the inference procedures used, can be thought of as a model of the expertise of the best practitioners of the field."[16] In terms of programming technology, the knowledge based approach has been described as an "evolutionary change with revolutionary consequences", [17] replacing the tradition of data + algorithm = program with a new architecture centred around a "knowledge base" and an "inference engine" so that: knowledge + inference = expert system. In fact, there are four essential components to a fully functional expert system: 1. the knowledge acquisition module; 2. the knowledge base; 3. the inference engine; and 4. the user interface. Knowledge acquisition is the process of extracting knowledge from experts. Given the difficulty involved in having experts articulate their "intuition" in terms of a systematic process of reasoning, this aspect is regarded as the main "bottleneck" [18] in expert systems development. The knowledge base stores information about the subject domain. However, this goes further than a passive collection of records in a database. Rather it contains symbolic representations of experts' knowledge, including definitions of domain terms, interconnections of component entities, and cause-effect relationships between these components. In legal expert systems this usually consists of formalised legal rules obtained from primary and secondary sources of law. Another layer of rules may also be obtained from less formal knowledge not found in published literature, [19] such as "practitioner's hand books and internal memoranda within legal practices". [20] These heuristics [21] add "experiential" to "academic" knowledge. [22] An inference engine consists of search and reasoning procedures to enable the system to find solutions, and, if necessary, provide justifications for its answers. The nature of this inference process is described in detail in Section 2. The user interface is critical to the commercial success of expert systems, particularly in the legal field, to enable lawyers with little or no expertise in programming, to gain access to the encoded knowledge. Typically this is in the form of prompting for information, and asking questions with "yes", "no" and "unknown" responses. Artificial intelligence, the foundation for legal expert systems, has run up against both practical and theoretical difficulties. While computers can beat the average human at "clever" tasks such as playing chess, they are "impossibly stupid" over tasks taken for granted such as speaking a language or walking across a room. [23] This casts doubt on whether many human activities do, as some artificial intelligence researchers suggest, consist of suppressed computational algorithms. There has been severe criticism by those who claim that knowledge, by its very nature, is not amenable to representation on a computer, [24] or that they achieve no more than simple competency. [25] Early misconceptions about the ease with which powerful and knowledgeable systems could be built for use by relative novices have given way to concern about real problems of knowledge elicitation and knowledge modelling. [26] Some opponents are convinced that the claims of artificial intelligence are exaggerated and their objectives unreachable. [27] Section 2 examines the nature of the inference engine, and suggests that its deductive procedures rest in pattern matching routines. It also explores issues of "fuzzy" and "deontic" logic. Section 3 explores the implications for knowledge representation, and questions whether devices such as "semantic networks", "frames" and "case based reasoning" are anything more than elaborate pattern matching constructs. Section 4 demonstrates that the need to be amenable to a deductive inference engine involves unacceptable distortion of law both at a practical and theoretical level. Section 5 argues that legal reasoning necessarily involves resort to social context and purpose, which is not tractable within current technology. Section 6 suggests that researchers ought to abandon legal expert systems, and instead concentrate on computerising more mechanical tasks such as legal retrieval and litigation support. A summary and conclusion is contained in Section 7. 2: PATTERN MATCHING AS THE CORE OF AN EXPERT SYSTEM The core of any expert legal system is its inference engine. This Section investigates the nature of computer inference at a basic level, and argues that it is little more than a pattern matching exercise. It is also argued that more sophisticated approaches, such as fuzzy logic, and deontic logic, are no more than extensions of the same principles. 2.1 _The Nature of Computer Inference_ Computer inference is undertaken by a simple strategy known as modus ponens. This means that the following syllogism is assumed to be correct: A is true (fact) If A is true then B is true (rule) \ B is true (conclusion) Computer deduction is obtained by conditioning the consecutive execution of instructions on matching, or failing to match, values in storage registers. The identical syllogism is obtained by a computer using a routine in the following terms: 1. check the value of register X1 2. compare the value of X1 to a value in register A 3. if X1 = A then: 4. change the value of register X2 to the value in register B The computer is conditioned by the value placed in register X1, either by the user or by satisfaction of some prior rule. A is taken to be true if the value of X1 is equal to a particular value in register A, representing some real world condition. The rule "if A then B" is contained implicitly within the structure of the routine by conditioning step 47 on the satisfaction of the condition X1 = A (i.e. A is true). The modus ponens is completed by step 47 which alters another register to equal a value B, thereby asserting that B is true. A programmer in BASIC or PASCAL, for instance, has some relationship in mind between the data supplied to the program and the output to be produced by computation. The input data are stored in the machine's memory, and the programmer's task is to devise a sequence of instructions to manipulate the data in accordance with the relationships she envisages. In such a case the inference engine constitutes the implicit algorithm contained within the sequence of instructions. Examples of such algorithmic knowledge based systems include Hellawell's CORPTAX, [28] CHOOSE, [29] and SEARCH, [30] all implemented in BASIC. 2.2 _Logic Programming_ Normal programming can at best maintain logical relationships implicitly within the program's structure. A number of researchers are now involved in logic programming, [31] which has been seen as the real technical breakthrough in this field. Some have extended logic programming to the point of writing expert systems by means of another logic based application, such as DARWIN. [32] Logic programming allows the programmer to specify logical relationships, not in terms of sequential instructions, but in terms of some symbolic language. [33] It is then up to the machine to compile the set of sequential instructions to maintain the desired relationship. A logic programming system can be regarded as a kind of rule-based system where the inference engine becomes a "mechanical theorem prover", [34] a machine for answering questions of the form: Do axioms (A0 . . . An) logically imply B ? The claimed advantages of rule-based logic systems over conventional programs are perspicuity and modularity. [35] Perspicuity is obtained by separating the rules (the knowledge base) from the logical operators (the inference engine). This has important implications for system maintenance and debugging. Modularity exists since the knowledge is split into small and independent rules. Legal rules, written in symbolic language, are manipulated through a process of "forward" and "backward chaining". A set of IF-THEN rules, constituting a "search space", [36] are compared against a set of facts to reach a logical conclusion. [37] In an expert system forward chaining simply involves matching the IF conditions to the facts, according to a predetermined order, which under the rules, dictate a conclusion. [38] Susskind describes this as a "control strategy" which "triggers" and "fires" the rules. [39] Backward chaining starts with the legal conclusion and searches for justifying antecedents in the knowledge base. In terms of programming this technique is more difficult since the search of the knowledge base is not along a single "path" but involves identification of all possible rules leading to the required conclusion. [40] In essence, it matches THEN variables with their IF antecedents and compiles a list of the paths thus generated. A "goal driven" expert system predetermines a conclusion and identifies the legal arguments and reasoning that can be used in support of that conclusion. [41] In logic programming terms this is no more than backward chaining across the search space. These processes of forward and backward chaining form the core of expert systems inference procedure. One notable feature of logic programming is the Horn clause, seen as a suitable extension to the "simple" predicate logic already outlined. It is of the form: A if B0 and . . . Bn where n >= 0. which consists of a single conclusion, A, and any number of conditions (B0, B1, . . .Bn). For example, the Horn clause: X is the father of Y if X is a parent of Y and X is male is a Horn clause with one conclusion and two conditions. Factual premises, such as "X is male" can be expressed as a Horn clause with one conclusion and no conditions. The significance of such a clause is that symbolic logic is not limited to IF-THEN statements, but may be extended to IF-AND-NOT-THEN statements. In terms of actual programming, however, the Horn clause is implemented as a bundle of IF-THEN assertions; each condition being checked separately for a pattern match, and the routine halting on failure to match. The pattern matching approach of logic programming is not relaxed but in fact tightened by the use of "integrity constraints", such as IF-THEN-ELSE structures, to close off the potential for negation by failure [42] and counterfactual conditional [43] difficulties. Constructed on the basis of Horn clauses, PROLOG and variations have been the platform for most logic programming projects, both in logic and procedure. APES, [44] implemented in PROLOG, is one widely used augmentation. [45] Using PROLOG as a symbolic logic structure involves rendering the domain knowledge in terms of Horn clauses, rewriting them in PROLOG syntax, and then executing the result as a program. PROLOG may itself provide a procedural basis for expert system platforms. Horn clauses may be backward chained as a procedure, working from conclusions to conditions and, as a sub-task, pattern matching each against its knowledge base, or user input. The program statements can thereby mix conditions which express legal rules with procedures to prompt the user for additional information. An example is Schlobohm's system to determine "constructive ownership of stock" under United States revenue laws. [46] The LEX [47] project is a more sophisticated application of the same principles. 2.3 _Fuzzy Logic_ Fuzzy logic is an attempt to escape the perceived inadequacy of binary logic. [48] Zadeh introduced the concept of the fuzzy set [49] to provide a formal way of speaking about imprecise concepts, such as "large" and "small". Rather than requiring precise values to be attached to particular characteristics, a spectrum of values, broken into categories, is used to match concepts, analogous to concepts in cognitive psychology. [50] The object of fuzzy logic is to convert continuous measurements into approximate discrete values. For example, a rule of the form: A PERSON IS A MINOR IF UNDER 18 YEARS OF AGE can be rendered by the following simple routine: 1. check register AGE 2. if AGE < 18 then: 3. set register PERSON to value 123 where 123 represents "minor". The spectrum of values "less than 18" is the fuzzy category. On a more complex level, matching can take place not only between ranges of values, but fuzzy sets. In binary logic, two concepts will be identical if and only if their membership functions, that is, their defining characteristics, exactly coincide. For instance, if F is a class of subsets of X, a set of characteristics defining legal concepts, then for Y and Z: Y is identical to Z iff fY(x) = fZ(x) for all x Rather than matching "identical" sets, fuzzy logic matches "closely identical" or "sufficiently close" sets. [51] To ascertain "closeness", a probabilistic metric is constructed. For example D(Y,Z) = Integral [{fy(x) - fz(x)}^2.p(x).dx] where p(x) is some probability distribution on X. D(Y,Z) is therefore a metric that depends on the choice of p(x). Using these definitions, one can test "closely identical" by inferring that: Y is identical to Z iff (1 - d(Y,Z)) >= d where d is some arbitrary threshold which can itself be used to trigger the operation of a rule. "Fuzzy logic" is therefore one way of rendering continuous or approximate concepts into terms amenable to computer deduction. It is, however, no more than an extension to logic programming techniques. Critics suggest that fuzzy logic is no more than oversophistication of arbitrary approximation; that its appearance of precision is spurious, and that its philosophical basis is uncertain when applied to legal concepts. [52] 2.4 _Deontic Logic_ In legal expert systems the nature of law as a normative system, [53] has given rise to a perceived necessity for incorporation of deontic logic. [54] Whereas traditional and classical logics provide formal canons for reasoning with empirical statements that have truth value, deontic logics provide standards for reasoning with statements which lack truth value [55] in the sense that they describe norms or imperatives. They cannot be characterised as true or false or logically related to each other or to statements of fact. [56] McCarty has consistently argued for "intuitionistic" rather than classical logic as the basis for representing legal concepts. He sets out some theoretical suggestions, as yet unimplemented, for the semantics of normative concepts in legal expert systems. [57] General foundations were laid by von Wright, [58] who developed a system of logic based on possibility and necessity. According to McCarty, "permission" exists in the union of all states and substates in which an action is necessarily true given the conditions of these states. This forms the "Grand Permitted Set" or a boundary condition for legality. [59] "Obligation" exists in the intersection of these sets. [60] In programming terms, "permission" entails backward chaining from the proposed action to all the states of the world. "Obligation" then is moving forward from all states of the world to find a common action. McCarty designed a language called LLD which allegedly possessed distinct advantages for legal applications in its use of action terms and deontic language. [61] However, his unimplemented proposal is problematic. LLD attempted to represent law in count terms, mass terms, states, events, actions, permissions and obligations. However, even McCarty admits that LLD failed to represent purpose, intention, knowledge and belief. [62] Jones demonstrates that a striking feature of McCarty's theorem is that an obligation to act did not logically imply its permission. In particular he demonstrates that under McCarty's analysis, one could logically derive "permission to poison the King from an obligation to serve him". [63] The only difference from logic programming in the suggested implementation of LLD lies in the use of fuzzy categories. That it does not stray too far from traditional logic programming is obvious since LLD is constructed almost entirely from Horn clauses. [64] McCarty therefore fails to tackle more difficult and fundamentally philosophical problems in deontic logic. [65] Stamper's LEGOL [66] project proposed a number of extensions to enable the system to handle concepts such as purpose, right, duty, judgment, privilege, and liability; [67] yet these were never implemented. [68] His latest project, NORMA, [69] has the object of relating all formalised symbols directly to the notions of agent, intention, and behaviour. However, it is doubtful whether this goal can be achieved, given that his languages are based in typical control structures such as sequencing of rules, if-then branches, and iteration, [70] hence easily rewritten as a logic program. [71] Sergot suggests that both Stamper's work and McCarty's LLD have simply taken standard semantics of logical formalism and presented their own variant. [72] Deontic logic may in any case be non-computational. Since the limit to current technology ultimately lies in the mechanistic linking of discrete relationships, the modelling of any "normative" aspect of law will not by its very nature be amenable to computer processing: "For there is not much sense in asking how . . . by having 255 in register 1234567 licences coming to have the number 128 in register 450254925." [73] Perhaps in light of this limit to technology, both the Oxford Project [74] and the Imperial College Group [75] have avoided deontic logic. Susskind reduced deontic logic to predicate logic by treating the normative aspect of law as merely linguistic. [76] Deontic labels were attached to different varieties of mechanical cause [77] and effect. [78] Normative statements were simply rewritten into declarative symbolic language. [79] The important implication from this work on deontic logic is recognition of the error in equating "logic" as understood by a computer with "logic" as understood in wider contexts. [80] In particular, reference can be had to MacCormick's distinction between "formal" and "everyday" logic, with the latter being based in common sense. [81] Researchers such as Stamper appear to be aware of such difficulties, but find themselves constrained by the existing technology. Whether legal reasoning can be computational is addressed in Section 5. 3: KNOWLEDGE REPRESENTATION AND THE PROBLEM OF CLASSIFICATION The previous Section demonstrated that the process of computer inference was limited to an elaborate process of pattern matching. This Section investigates the implications for knowledge representation; in particular, that it is necessary for knowledge to be represented in terms of IF-THEN rules. It is also argued that more "sophisticated" representation techniques, such as "semantic networks" are no more than elaborations of this basic structure. 3.1 _Pattern Matching and Open Texture_ To be implementable, the knowledge base must be structured so as to be amenable to deductive inference procedures. In theory, it is possible to use any form of symbolic logic as the representational formalism as long as it is appropriate to a deductive inference engine. This condition requires that the knowledge base must be in the form of pattern matching rules. In logic programming, computation is deduction, and the task of the programmer is therefore to pose a problem suitable for a deductive process. [82] The major difficulty encountered is what broadly may be termed "semantic indeterminacy". [83] Not all legal rules are appropriate for application in all situations. [84] Legal expert systems have been acknowledged to be only capable of solving problems referred to as "clear cases of the expert domain". [85] Yet what is a clear case? The Oxford Project defined a clear or "easy" case as one easily solved by an expert, yet hopelessly difficult for non-experts. [86] Gardner [87] drew the distinction between hard and easy cases by describing the latter as situations whose verdict would not be disputed by knowledgable and rational lawyers, whereas they may rationally disagree as to the former. The real answer for legal expert systems lies in the nature of the computation process. When presented with the facts of a case, the expert system must decide whether or not a rule applies. Since "fact situations do not await us neatly labelled, creased, and folded" [88] the difficulty lies in subsuming particular instances under a general rule. [89] A "hard case" is therefore one where the system fails to match the appropriate pattern, thereby preventing a rule from firing. As computer logic relies on pattern matching, knowledge representation necessarily must encounter problems of classification. [90] What is "ultimately beyond the grasp of a computer," states Detmold, "is not complexity, but particulars". [91] This difficulty is often termed "open texture". The notion of open texture is obtained from Hart's analysis. In the now infamous regulation: NO VEHICLES ARE PERMITTED IN THE PARK the open-textured term here is vehicle. The difficulty in terms of legal expert systems is how the program can classify an object as being a "vehicle" falling under the rule. Hart suggests that general words like "vehicle" must have a set of standard instances in which no doubts are felt about its application. There must be a "core of settled meaning", but there will be, as well a "penumbra of debatable cases". [92] In terms of computation, a case is within the core of settled meaning, and is classified as "easy" where there is a matching pattern in the knowledge base. Cases in the penumbra of doubt are hard, since they cannot be classified by the system. The difficulty for legal expert systems, then, is to build a system for classification, so that the pattern matching process can take place. Skalak [93] suggests there are three theoretical models of classification: - the classical model - the probabilistic model - the exemplar model All three models have been extensively used in expert systems technology. The following analysis demonstrates that the first two models have little to distinguish them in practical effect, and together constitute an inadequate response to the problem of open texture. The exemplar model has been used as justification for case based reasoning approaches, but the term has been misused, and such cases are argued instead to fall into the "probabilistic" model. The exemplar model is returned to in Section 6, where it is used as the basis for suggested development of computer applications to law. 3.2 _The Classical Model_ In the classical model, a concept is defined by necessary and sufficient conditions. Hafner [94] suggests that these conditions can be formally represented by knowledge structures involving decision rule hierarchies, taxonomic hierarchies, or role structures. Decision rule hierarchies specify the conditions under which a concept is true or false. "Vehicle" may be defined by a set of characteristics such as "four wheels", "engine" and so on. In programming terms, this means that the IF antecedents are themselves THEN consequents based on sets of prior conditions which constitute the "definition" of a term. This quickly builds into a "decision tree" structure. [95] Statute law, particularly statutory definitions, are seen particularly suitable for rendering into what amounts to typical Horn clauses. [96] A prominent example of this technique is the modelling of the British Nationality Act 1981. [97] Others include the United Kingdom supplementary benefits legislation, [98] and STATUTE, now used by some Australian government departments. [99] Taxonomic hierarchies define sub-types of concepts, placing different objects into groups and sub-groups. For instance, the class "vehicle" may have among its sub-classes "car", which in turn may be further sub-classed into "Toyota", "sedan" and specific instances based on model types, years, and so on. [100] Any taxonomic hierarchy can, however, be represented in terms of a chain of decision rule hierarchies, or IF-THEN rules. Role structures are modelled in "frames", and "semantic networks". A semantic network [101] is a collection of objects called "nodes". These are connected by "links". [102] Typical links include "is a" links to represent class-instance relationships and "has a" links to represent part-subpart relationships. Interconnected, these may quickly build into a complex web of relationships. A frame [103] is a subset of a semantic network, being a representation of a single object with a set of "slots" for the value of each property associated with the object. All links in both semantic nets and frames are, however, functionally equivalent to taxonomic hierarchies. Hayes demonstrates that both semantic networks and frames are no more than elaborate logic programs, and concludes that they hold no new insights. [104] The semantic network may be forward and backward chained as a set of logical rules, just as would a rendering of a set of Horn clauses in PROLOG. [105] For instance, in McCarty's TAXMAN project [106] the domain was modelled in terms of objects such as corporations, individuals, stocks, shares, transactions &c. Each object is described by a "template", being a collection of the object's properties, such as name, address, size, and value. These properties are then linked and indexed. Each "bundle of assertions" constitutes the object's "frame". [107] These structures are aimed at answering questions such as: "Does the taxpayer and her family have a controlling interest in the stock of a company which is a partner in a partnership which owns an interest in XYZ Ltd?" [108] In TAXMAN 2 McCarty proposed a more elaborate semantic net based on a "prototype-plus-deformation" model. Essentially this sets one frame as being the default for each class of object, with incremental modifications to slots based upon fuzzy categories. Unfortunately the concept was never implemented. As a result, McCarty offer no solution to algorithmic issues such as how to choose, index, and search the space of prototypes, and their relationships to actual cases. [109] 3.2 _The Probabilistic Model_ Some argue that legal concepts cannot be adequately represented by definitions that state necessary and sufficient conditions. Instead legal concepts are incurably open-textured. [110] Typically an expert system associates some kind of "certainty factor" with every rule in its knowledge base, obtained from probabilities, or fuzzy logic, or some combination of the two. [111] Firstly, probabilities are used alongside facts and rules, as a "slot" in the knowledge base. To each fact and rule is attached a certainty factor between zero and one. Concepts are mechanically linked, but the final output includes a composite probability. For example: A is true (0.8 chance) If A is true then B is true (0.75 chance) \ B is true (0.6 chance = 0.8 x 0.75) Secondly, fuzzy logic is called into play when classification of facts involves weighting particular features. In the former the rule "fires", but a certainty level is attached to each fact and rule. In the latter, the rule will only fire at defined threshold certainties. If the weighted average of a set of characteristics add to a threshold amount, the facts are classified accordingly. 3.3 _Case Based Reasoning_ Using precedents by induction and analogy is seen as advantageous in overcoming apparently intractable problems of classification. [112] However, both analogy and induction are inherently non-computational. [113] Case based reasoning is one attempt to imitate these techniques, allegedly based on an exemplar model. In the exemplar model, the user is presented with prototypical instances, or "mental images", on which to base her classification. This approach differs significantly from the two previous models in that it is primarily designed to leave the task of classification to the user. Most case based legal expert systems instead use a database of examples linked to the decision given in particular cases. When presented with a new case for decision the system will attempt to match the case under consideration with the examples, either stereotypical [114] or actual, in its database to extract those which appear to be most similar. On that basis it will attempt to predict the outcome of the new case. For instance, Popple's SHYSTER [115] applies rules until the meaning of some open texture concept is required. At this point a case based reasoning mechanism attempts to resolve this uncertainty. [116] A matching algorithm is used to measure the similarity of cases in terms of "case features". Each case is modelled as a frame with significant features contained in "slots". [117] These features are then weighted by some statistical method. [118] The object of constructing these "similarity metrics" [119] is to retrieve the most "on-point" cases. [120] Using weighted characteristics is described as a "dimensional" [121] approach, such as Betzer's "3-D" system which uses relative weights in a "procedure sweep" to fill in "gaps" in the knowledge base. [122] In addition to the facts, the cases themselves are often weighted in a manner "meaningful to lawyers", usually to reflect some sense of stare decisis. For instance, the weights might be determined by the level of the tribunal. [123] Although "case based" reasoning allegedly possesses an advantage over rule based systems by the elimination of complex semantic networks, [124] it suffers from intractable theoretical obstacles. Apart from the question of choice of a matching algorithm, without some further theory it cannot be predicted what features of a case will turn out to be relevant. Too often, "legally significant parameters" [125] are facts deemed important by the programmers, [126] with no grounding in any articulated theory, even though the utility of such systems depends critically on the set of attributes selected. [127] Both selection of attributes and the choice of associated weights are therefore highly arbitrary. [128] On this analysis, case based reasoning constitutes an extension of the probabilistic model rather than a true exemplar model, in which the task of classification is left to the user. The potential of this latter model for building applications is examined in Section 5. 4: PHILOSOPHICAL IMPLICATIONS OF THE DEDUCTIVE INFERENCE ENGINE The previous Sections have demonstrated that the task of knowledge representation was to provide a symbolic representation of knowledge in a form amenable to the deductive inference engine. The primary reference point was logic programming, that is, formalisation of the law into a set of Horn clauses. It was also argued that techniques such as "fuzzy logic" "semantic networks" and "case based reasoning" are no more than elaborations of logic programming, [129] and not, as some would argue, "second generation" systems going beyond deductive inference. [130] This Section carries this analysis beyond the practical and into the philosophical. It has been thought inescapable that a legal expert system which attempts to emulate the reasoning processes of a lawyer must embody theories of law that must in turn rest on more basic philosophical assumptions. [131] Building a legal expert system is thus described as not being just an exercise in computer programming, but requires "solid and articulated" jurisprudential foundations. [132] Researchers in this field appear, however, to have discounted or ignored the value of close analysis of the field's theoretical assumptions. This Section demonstrates how, to avoid theoretical obstacles, the nature of law, its epistemological basis, and the task of jurisprudence have all been subjected to unacceptable distortion. 4.1 _Isomorphic Representation or Distortion?_ The activity of legal knowledge representation is said to involve the operation of interpretative processes whereby the formal sources of part of a legal system are scrutinised and analysed, so as to be both faithful in meaning to the original source materials, and in a form which is computer encodeable. This principle is termed "isomorphism". Doubts have been raised as to whether it is possible to meet both objectives. It is said that the knowledge engineer must desist from imposing her own interpretations, lest she be universally condemned for misrepresenting the law. [133] Yet it is difficult to reconcile Susskind's claim of isomorphism with his admission that the process is in fact one of complete "reformulation" or "rational reconstruction". [134] Although isomorphism requires the formalisation of rules to be sufficiently expressive to capture their original meaning, [135] Levesque and Brachman have demonstrated that there is a significant trade off between the expressiveness of a system of knowledge representation and its computational tractability. [136] Susskind admits that it is not possible, without "extensive modification and inconvenience" to accommodate legal knowledge within the restrictive frameworks offered by currently available computer programming environments. [137] Moles provides a typical example of the modelling of British coal insurance claims. [138] This involved taking statutes and cases and then "translating" them into six different structures using three separate applications into the target representation language. After being "translated, cut up into bits, precised, further analysed into [frames], which are then stored in another structure", he suggests it would be a "miracle" if they were "isomorphic" to the original texts. [139] It would appear that terms such as "isomorphism" may be no more than "syntactic sugar" [140] used to "sweeten" the acceptability of what must be a distorting process. 4.2 _Law as a System of Easily Interpreted Rules_ Statutory interpretation has been predominantly characterised as involving a literal interpretation, [141] particularly in tax law, [142] allegedly idiosyncratic in being construed both literally and strictly. [143] In case law, legal sources cannot be as easily "formalised" or "normalised", [144] but must to some degree be interpreted. However, Susskind takes a dangerous step in suggesting that the task of the knowledge engineer is to "sift the authoritative ratio decidendi from the text eliminating obiter dicta and other "extraneous" [145] material. Moreover, he argues that this can be easily extracted, not by a thorough examination of the case but by reading the headnote alone. [146] Representation of cases in knowledge bases typically are compressed into a "headnote" style, including citation, court, date, facts, and holdings. [147] Cost may also be a factor behind this characterisation of law. Susskind suggests that knowledge engineers need only avail themselves of the services of the legal expert to "tune" the knowledge base. [148] This view, however, that the law is a formal rule-governed process, ignores a great deal of learning stretching back over a century - and more recently in the form of critical legal studies - arguing that the law is far from determinate. [149] The law is at least an "elastic" phenomenon in which students have traditionally been taught and encouraged to "flip" legal argument. [150] The conception of legal decision-making as a formal rule-governed process has been eroded by a judicial move towards "realist" scepticism of rigid rule structures. For example, members of the Australian High Court have indicated a rejection of formalism and adoption of a more active assessment of legal principles with respect to justice, fairness, and practical efficacy. [151] Advocates of case based reasoning attempt to accommodate realist criticism by suggesting that fact patterns can explain legal decisions independently of any "surface discourse" of law. [152] The critical assumption is that judges decide even hard cases in a rule based manner. Levi [153] supports this view in arguing that legal reasoning, while not being purely a system of applying the law mechanically to facts, does embody rules obtained by analysing the similarities and differences in decided cases. Such researchers argue analogously to theorists, such as Goodhart, who suggest examination ought to be focussed on the facts treated as material and immaterial by the court. [154] Stone, however, argues that there is a critical distinction between the ratio which explains the decision and the one which binds future courts. More often than not, the critical facts are those treated as material by the later court, and even if they are identifiable, they can be stated at multiple levels of generality. [155] Other legal systems, particularly in some parts of Europe, may be more suited to this characterisation of law. For example, in the Scandinavian legal system, one overall guiding principle is the prohibition of decisions which are non liquet [156], which is considered a serious fault. [157] 4.3 _Change_ A severe impediment to the routine use of knowledge based technology for practical legal applications lies in the unresolved problems associated with the "maintenance" of such systems, that is, how to continually update the system with primary sources. [158] One group describe how after exhaustively studying over 1,000 cases under the Canadian Unemployment Insurance Act, it was amended in 1990 rendering their work irrelevant. [159] Most approaches are inadequate, either for expressly assuming a constant state of the law, or avoiding primary sources and instead modelling directly the heuristics of the expert. [160] The logic programming approach of the Imperial College group, whereby the expert system is formalised to correspond to individual sections of a statute, is argued to be easily modifiable. However, a fully functioning expert system requires a layer of pertinent heuristic knowledge to avoid a "layman's reading" of an Act. [161] Once the formalisation is structured, explained and augmented in this way, modifying the system is no longer straightforward. Schlobohm suggests that, as a result, human experts would have to modify the heuristic rules whenever the law changes, and the entire system containing the new rules would then have to be debugged. [162] Similarly, use of modular approaches such as the Chomexpert system have proved inadequate. [163] It is difficult to encode statutory rules at even the most basic level without making inappropriate commitments as to how they will be interpreted in future. [164] 4.4 _Epistemology of Law_ Susskind suggests that law is not an abstract system of concepts and entities distinct from the "marks on paper" that are the material symbols of it. [165] The difference between legal expert systems and scientific systems such as PROSPECTOR and MYCIN lies, according to Susskind, in that scientific laws are to be "discovered" in the empirical world in general, while legal rules can be extracted, as an acontextual linguistic exercise, from scrutiny of formal legal sources. Under this analysis, knowledge engineers need go no further than the written text, hence Susskind argues that the "bottleneck" of knowledge acquisition is effectively dissolved. This is a dangerously narrow epistemology to adopt, [166] since researchers in this area do not sufficiently distinguish between the writing, and the meaning of the writing. [167] In Section 5 it is argued that meaning can only be found in a social context. 4.5 _The Nature of Jurisprudence_ Although the foregoing suggests that many theories in jurisprudence conflict significantly with important assumptions of expert systems technology, many of these fundamental theoretical difficulties have been downplayed or eliminated. When faced with theories which imply, for instance, that there is no future for expert systems, some researchers have expressly rejected the usefulness of jurisprudence. [168] Critical legal theory is therefore characterised as "unacceptable". [169] Even if jurisprudence is wholly ignored by knowledge engineers, they suggest that the only risk is that the systems they design might be of some "inferior quality". [170] Others "rationally reconstruct" jurisprudence into an acceptable form. For instance, Susskind asserts that the activity of any "legal science" is to impose order over unstructured and complex law by recasting it into a body of structured, interconnected, coherent, and simple rules. [171] Smith and Deedman go further and argue that the task is to transform apparent indeterminacy into a completely rule-governed structure. [172] The same can be said for the portrayal of the epistemology of jurisprudence. Just as "complex" law is recast into "simple" rules, the task of Susskind was to take "confused and perplexed" jurisprudence, and obtain "consensus" over relevant issues. What Susskind does to find "consensus" in legal theory, is to allegedly statistically sample the literature. [173] However, the "sample" was limited firstly to works of analytical jurisprudence, and secondly to British writings from the mid-1950's. [174] Perhaps unsurprisingly, the influence of H.L.A. Hart's concept of law as a system of rules was overwhelming. As an adjunct, Susskind further asserted that to be "jurisprudentially impartial", that is, to embody no "contentious" theory of law, an expert system must reason only with rules. Any facility for reasoning with non-rule standards [175] was rejected out of hand. A significant internal inconsistency emerges when it is appreciated that Susskind believed that it was sufficient justification for use of rules that this "consensus" identifies legal rules as necessary but insufficient for legal reasoning. [176] Perhaps the clue to why these works were chosen lies in the fact that they constituted "the source materials with greatest potential given the overall purpose of the project". Susskind notes that his work was intended to "eliminate much of the future need for extensive scrutiny of non-computationally oriented contemporary legal theory". [177] Here the inference engine is most clearly "driving" jurisprudence. 4.6 _Jurisprudence Turned on its Head_ Niblett claimed that "a successful expert system is likely to contribute more to jurisprudence than the other way around". [178] If the suggestions of researchers such as Susskind are taken seriously, they turn jurisprudence on its head. Theory is not used as a basis for practice, but instead implementability in technology is used as the touchstone for accepting the truth or falsity of the theory. A particular feature of artificial intelligence literature is that its rigour lies not in experimental corroboration, or any theoretical soundness, but implementability. [179] Hofstader [180] suggests that so long as the artificial researcher takes care to construct theories which can be written down as a sequence of algorithmic or computational steps, these theories can be implemented, thereby "confirming" the theories underlying the process. Implementability per se leads to a self-perpetuating methodology: since an artificial intelligence researcher will use concepts of computational theory to construct theories, it is necessarily implementable. Legal expert systems researchers fall into this model by rejecting "unacceptable" legal theories, and reformulating the remainder in computational terms, to eliminate potential obstacles to the prosperity of their research programme. This abandonment of serious inquiry into jurisprudence by researchers into legal expert systems may give credence to Kowalski's fears that the field may have cut itself off as a specialist discipline and established its parameters prematurely. [181] Brown notes that at a 1991 conference, few if any papers questioned the basic assumptions of the field. [182] Niblett claimed that "a successful expert system is likely to contribute more to jurisprudence than the other way around". [183] The foregoing demonstrate that these words ring true. Law and jurisprudence, to form an acceptable basis for expert systems research, has been reformulated in computational terms, to eliminate philosophical "technicalities". [184] 4.7 _Failure to Recognise Limitations_ Leith has argued for a rejection of legal expert systems on the basis that they simplify the law to such an unacceptable extent that they have little or no value in legal analysis. [185] Yet while some engineers of legal expert systems may be fully aware of the limitations already discussed, it is not necessarily the case that other researchers, and more importantly, the users of these programs will also be so mindful. This article agrees with Leith's implied suggestion that many accounts of work in this area refuse to acknowledge that there are significant limitations. For example, McCarty felt able to say: "[Law] seems to be an ideal candidate for an artificial intelligence approach: the "facts" would be represented in a lower level semantic network, perhaps; the "law" would be represented in a higher level semantic description; and the process of legal analysis would be represented in a pattern-matching routine." [186] Susskind has, however, admitted that expert systems might not be amenable to corporate, commercial and tax law, but would be apposite, for example, but to limited instances such as the Scottish law relating to liability for damage caused by animals. [187] Such limitations, often given little attention, should be made clear, and "plausibility tricks" avoided. [188] There is a very real danger that users will significantly overestimate the value of the analysis they obtain from such a program, particularly in light of the wealth of optimistic literature and when it is described as "expert". 5: LEGAL REASONING AS AN INTRACTABLY COMPLEX SYSTEM The previous Section demonstrated how law and jurisprudence have been unacceptably distorted to be amenable to expert systems technology. Moles suggested that researchers have deliberately ignored fundamental problems since they were committed to the use of a "particular computing tool", and not to the understanding of law. [189] This article identifies this tool as the inference engine itself. The following section addresses the non-computational nature of legal inference. 5.1 _Search for a Deep Model_ There has been a growing trend in legal expert systems to speak of "deep knowledge" or "conceptual knowledge" as something distinct and preferable to "shallow" knowledge. [190] McCarty calls for the development in law of "deep" systems akin to CASNET [191] in which the disease is represented as a dynamic process. [192] The depth of a system has been described as the extent to which programmes contain not only rules for mapping conclusions onto input scenarios, but also a representation of the underlying causes. [193] The Imperial College Group suggest that deep structure in legislation is the isomorphism to that legislation, on the basis that each Horn clause represents some clause in the legislation. [194] In addition, case based reasoning has been described as employing a "deep structure". [195] The advantage stemming from both of these descriptions is that they cast deep structure into computational terms. [196] This is another example of technology driving the underlying theory. On the other hand, McCarty argues that resolution of the difficulties of open texture are related to a sense of "conceptual coherence". [197] In addition, while theoretical approaches are emerging to cope with problems of legal change, [198] a unifying theme is a striving for an undefined "normative enrichment". [199] This Section argues that deep structure is to be found in social context and purpose, which are non-computational. 5.2 _Interpretation in a Social Context of Shared Understanding_ Law is not, as legal expert systems would portray it, self-contained and autonomous, [200] but in fact is embedded in social and political context. That legal concepts draw upon ordinary human experience is precisely what makes them so difficult for an artificial intelligence system. [201] Whenever human behaviour is analysed in terms of rules, it must always contain a ceteris paribus condition; in essence referring to the background of shared social practices, interests and feelings. Even if we accept Susskind's characterisation of law's ontology as going no further than the "marks on paper", semantic problems will still arise since these marks are not created in a vacuum, but are the result of purposive social interaction, and must be so interpreted. [202] Using one recognised example, the injunction: DOGS MUST BE CARRIED ON THE ESCALATOR can only be interpreted based on the understandings, for instance, that a dog's small feet may become trapped in slots and moving parts; that humans generally feel some concern for dogs, and therefore do not wish to see them "mangled". Thus an adequate interpretation of any rule requires that we locate it in a complex body of assumptions. [203] Minsky noted that intelligent behaviour presupposes a background of cultural practices and institutions which must be modelled if computer representations are to have any meaning. [204] Wittgenstein's arguments that the meaning of language must be based in social use and a community of users are worth rereading in the light of Searle's Chinese room analogy. [205] How can the computer have this sort of direct access to language? [206] Kowalski and Sergot admit that a computer must operate by "blind" and "mechanical" application of its internal rules. [207] 5.3 _Open Texture as an Intractable Problem_ If legal reasoning was really some "pointing" [208] or "cataloguing" [209] procedure, Hart's suggestion that the task of legal institutions is to approach greater refinement in definition by adjudicating [210] on particular cases, may be attractive. [211] Open texture may then be marginalised by a progressive refinement of categories; in computational terms, weaving a more elaborate semantic net. To model social context in a knowledge base, however, may be an impossible task. Popper demonstrates that context entirely depends on point of view. [212] Harris suggests that any view of the law must be a phenomenological one which takes account of shifting foci of interest. [213] The difficulty is that a great deal of social context will not be "conscious" and expressible, but will constitute a hidden set of assumptions on which human decisions will be based. It is impossible to focus attention onto elements of that context without creating a new subconscious context. Polyani describes this as the difference between focal and subsidiary awareness. [214] As Berry [215] demonstrates, if people learn to perform tasks so that important aspects of their knowledge are implicit in nature, then knowledge engineers will be unable to extract this knowledge and represent it in a meaningful way in an expert system. [216] Husserl, for instance, discovered that construction of even simple "frames" involved coping with an ever expanding "outer horizon" of knowledge. He sadly concluded at the age of 75 that he was a "perpetual beginner" engaged in an "infinite task". [217] This is a fundamental difficulty with artificial intelligence in all its applications. [218] 5.4 _Purposive Interpretation and Intention_ Hempel [219] argued that ad hoc modifications to a theory were limited by the increased complexity of the theory and that, after a certain threshold level of complexity was exceeded, scientists would naturally and logically pursue simpler alternative theories. [220] Here we may learn from science. Certain physical and chemical systems have been discovered that display uncanny qualities of co-operation, or organise themselves spontaneously and unpredictably into complex forms. These systems are still subject to physical laws, but laws that permit a more flexible and innovative type of behaviour than the old mechanistic view of nature ever suggested. The lesson from chaos theory is that seemingly complex systems can be defined in terms of simple but not mathematically tractable models. [221] Legal reasoning is not mechanical. [222] Social context and shared understandings can be dealt with in terms of the simple, elegant, but non-computational model of purposive interpretation. Searle's Chinese room analogy identifies intentionality as the benchmark of the mental, and refutes claims that intentional mental predicates, such as meaning, understanding, planning, and inferring, can be attributed to a mathematical computational system. [223] Susskind prefers to avoid purposive theories, [224] since such theories imply that law is not simply a question of linguistic pattern matching but instead involves examination of social practices and human intentionality. [225] Similarly, case based reasoning is seen as a way around having to tackle "full blown" statutory interpretation involving legislative intent. [226] Law is a practical enterprise, concerned to guide, influence or control the actions of citizens. Since any action is purposive, any philosophy of action must be a philosophy of purposes. [227] When a court applies, say, the statutory term of our example, "vehicle", to a particular contraption, the meaning of "vehicle" is found in an analysis not only of the purpose of the law, but of the purpose for which the vehicle was to be used. [228] For example, Fuller responded to Hart in these terms: "What would Professor Hart say if some local patriots wanted to mount on a pedestal in the park a truck used in World War 2, while other citizens, regarding the proposed memorial as an eye-sore, support their stand by the "no vehicle" rule? Does this truck, in perfect working order, fall within the core or the penumbra?" [229] One could make similar arguments when a "NO DOGS ALLOWED" sign confronts a seeing eye dog, or one that is stuffed or anaesthetised. [230] It is difficult to reconcile Hart's acontextual approach to legal interpretation with his own view of actors within the legal system holding an internal normative view of the rules. [231] Following a rule equals "obeying the law" only where a purposive personal commitment has been made to a rule structure. [232] Applying modern literary and linguistic theory to the law, [233] some suggest that no text has meaning without the active participation of the reader, [234] and an "interpretive community" of which the reader is a part. [235] The use of figurative language, imagery and metaphor is integral to legal discourse. [236] Ideological symbolism is inescapable. [237] What counts as the relevant facts depends entirely on context, [238] and cannot be determined by programmers ex ante. [239] Language is the very condition of intention, and intention is the vehicle of meaning. [240] 5.5 _Humanistic Implications_ The implication of the foregoing suggests that the law cannot be amenable to a legal expert system, as this involves denying social context, purpose, and essentially humanity. A humanistic critique would argue if expert systems have any degree of success in modelling "the law", the result would be "profoundly humiliating". [241] Weizenbaum stated that if artificial intelligence fulfils its promises then this implies that man is merely a machine. [242] In similar vein, the success of legal expert systems might imply that the law itself is a machine, and that lawyers, perhaps even judges, can be replaced by computers. 6: THE WAY FORWARD The preceding sections have demonstrated that the use of a deductive inference mechanism, and the consequent need for knowledge represented to be amenable to such an engine, will lead to unacceptable distortion of both the law, its philosophical underpinnings, and its humanity. How are lawyers then effectively to utilise the information technology resource? This article adopts the basic message in Tapper's insightful 1963 piece. [243] The range of activities to which computers ought be used must be limited to activities which can be reproduced by the machines. Tapper tentatively describes the distinction as one between "mechanical" and "creative" tasks. [244] If the argument of this article is accepted, the way forward involves relocation of the inference engine from the computer to the human user. This section explores possibilities for "decision support systems", which presents material to the user on which she alone performs the specifically legal reasoning. [245] 6.1 _Decision Support Systems_ Recall that in the exemplar model, the user is presented with prototypical instances, or "mental images". This approach differs from the other models of classification in that it is primarily designed to leave the task of classification to the user of the system. The reason why the user, rather than the machine, ought perform the legal inference is that legal reasoning is non-computational, as Section 5 has demonstrated. Despite growing recognition that research perhaps ought to be oriented towards "decision support systems", such systems have been designed to first reason with the legal data and then present such reasoning to the user to support her conclusion. [246] This approach is hazardous since it may predetermine the human conclusion to a large degree. [247] To dispute the computer inference the user would require knowledge of the area of law to a degree where the computer would not need to be have been consulted in the first instance. [248] Decision support systems, then, differ significantly from expert systems in that the heart of the problem - the inference engine - is relocated in the user of the system. Computers should then be utilised for mechanical and time-consuming tasks for which they are best suited. In particular, this Section suggests three significant uses: - structured legal information retrieval; - calculation based on strategies; and - litigation support and "legal econometric" systems. 6.2 _Legal Information Retrieval_ Firstly, searching for primary and secondary legal sources is a costly and to a significant degree a mechanical exercise. Efficient retrieval of legal information is vitally important. Tapper suggested in 1963 that lack of resources to those operating outside provincial centres, and concentration of materials within large organisations was productive of injustice in favour of powerful sections of the community. [249] Modern statute and case databases have gone some way to addressing this problem. Generally, systems such as LEXIS, SCALE, and INFO1 use Boolean keyword search routines, [250] but these have obvious disadvantages. [251] Some limited advances have been made with, for instance, "Hypertext" cross-referencing, [252] and "probabilistic" elaboration of keyword searches. [253] It has long been assumed that retrieval based on the meaning and content of documents, and indexed in terms of legal concepts, [254] would be far more appropriate. [255] A variety of techniques have emerged for indexing, including use of discrimination trees, [256] and explanation based generalisation. [257] Research is progressing towards a "hybrid" approach of linking case databases with statutory material and legal texts. [258] Hafner [259] has constructed a database on United States negotiable instruments law designed to retrieve cases based on typical problems which arise in legal disputes. McCarty has suggested that it would be more fruitful to look at legal argument than to develop a theory of correct legal decisions. [260] Similarly Bench-Capon and Sergot suggest that open texture should be handled by giving the user for and against arguments in borderline cases. If so, a computer system will be concerned, not with the production of a conclusion, but rather with presenting the arguments on which the user may base her own conclusions. [261] On this basis, Ashley and Rissland have designed a system called HYPO, [262] which emerged from Rissland's earlier work on reasoning by examples. [263] It does not use an inference engine for legal analysis but instead aims for conceptual retrieval based on structure of legal argument. [264] The system's inference engine is used for some statistical processes used to decide which primary materials to retrieve. The cases relevant to the issues identified by the user are retrieved and arranged in terms of argument for and against a decision in a new case. [265] The system is further supplemented by a set of "hypothetical" cases. [266] The actual legal inference on the basis of the material retrieved is left to the user of the system, which distinguishes HYPO and similar text retrieval systems from many of the case based reasoning systems earlier. 6.3 _Calculations and Planning_ A second application would be to utilise the mathematical functions of computer systems. The computer performs inference, but essentially calculates outcomes based on strategies already formulated by an expert who has himself interpreted the legal materials. Michaelson's TAXADVISOR [267] is one example. It calculates tax planning strategies for large estates, based on strategies obtained from lawyers experienced in tax advice. There is little more legal inference in this than calculating a share portfolio to maximise return based on a broker's personal model. Similarly, systems have been suggested which will assist in financial planning, for instance by forecasting retirement pensions. [268] 6.4 _Litigation Support and Jurimetrics_ Finally, a decision support system may more clearly focus on litigation strategies. These may be developed with the assistance of expertise, or by techniques such as hypothesis and experiment. [269] Such systems do in fact make inferences, but these are not inferences of law, but inferences based on strategies already defined by expertise or essentially what amounts to empirical research. In that sense, the inference procedures are extensions to the calculation and planning examples. One example is the LDS [270] system, implemented in ROSIE [271] by Waterman and Peterson. It advises on whether to settle product liability cases, and an advisable amount, based on factors such as abilities of the lawyers, characteristics of the parties, timing of claim, type of loss suffered, and the probability of establishing liability. The primary goal of LDS was not to model the law per se but rather the actual decision making processes of lawyers and claims adjusters in product liability litigation. Another example is SAL, [272] intended to advise on an appropriate sum to settle asbestos injury claims. In such systems, the computer is modelling non-legal factors which may influence the outcome of a case, in order to assist the lawyer in deciding her ultimate strategy. In Australia the Government Insurance Office has developed COLOSSUS, a sophisticated system to detect possible fraudulent personal injury claims, and tag them for investigation by its officers. [273] Similar systems have also been suggested as aids not only in litigation, but dispute resolution strategies. [274] The information contained in such systems may as an adjunct constitute an important resource for sociological study, such as Bain's modelling of subjective decisions of judges of particular varieties of crime in the United States. [275] In this case, the expert system constitutes jurimetrics, a legal version of econometrics. 7: SUMMARY AND CONCLUSIONS Computerisation of the legal office will continue, but the message from this article is that researchers must be acutely aware of the philosophical underpinnings of their work. In particular, the usefulness of legal expert systems is severely questioned. Use of such systems has involved an unacceptable level of distortion both of the nature of law and of jurisprudence. This is not a case of "carbon", [276] "biological", [277] or even "neural" [278] chauvinism, but a demonstration that expert systems technology have made a poor choice of domain in law. Blame has been laid for such distortion on the core of the expert system: the pattern matching inference engine. Legal inference, on the other hand, relies on purpose and social context, implying that computational models of sufficient richness are not tractable. This article suggests that, given current limitations of computer technology, the quest for an artificially intelligent legal adviser is misguided. In the future, however, these limitations may be overcome. For example, work being undertaken in parallel distributed processing is producing significant results with respect to low level "intelligent" processes, including perception, language, and motor controls. This is based on the assumption that intelligence emerges from interactions of large numbers of simple processing units, and represents a significant break away from increasingly complex rule-based structures. [279] While this article cannot address such possibilities within future technology, it is suggested that the basic pattern matching /rule-governed principles will limit computers for some time. It is therefore suggested that researchers instead investigate decision support systems as a more useful alternative. Relocation of the inference engine will mean that knowledge representation will no longer need be amenable to computational inference, but human inference. The computer's inference engine should instead be used instead for searching procedures, and computation. Some possibilities have been noted. Ardent advocates such as Tyree suggest that despite their difficulties, legal expert systems are a cost-effective second-best solution. The choice is portrayed not between human advice and machine advice, but in an era of high costs of justice, between machine advice and no advice at all. [280] While economic factors are important, [281] humanistic factors must not be forgotten. Law plays an important role in modern civilisation. It must maintain a close relationship with the social and political forces shaping society, and not merely regress into a "technology", a tool to be used by competing social forces. [282] --------------------------------------------------- ENDNOTES * J Searle, "Minds, Brains and Programs" (1980) 3 Behavioural and Brain Sciences 417. 1 For examples see NJ Bellord, Computers for Lawyers (Sinclair Browne: London, 1983); and T Ruoff, The Solicitor and the Automated Office (Sweet & Maxwell: London, 1984). 2 q.v. J Bing (ed.), Handbook of Legal Information Retrieval (North-Holland: Amsterdam, 1984). 3 See Section VI, infra. 4 Inferring molecular structure from mass spectroscopy data; q.v. RK Lindsay, BG Buchanan, & J Lederberg, Applications of Artificial Intelligence for Chemical Inference: The DENDRAL Project (McGraw-Hill: New York, 1980). 5 Advising on the location of ore deposits given geological data; q.v. RO Duda & R Reboh, "AI and Decision Making: The PROSPECTOR Experience" in W Reitman, Artificial Intelligence Applications for Business (Ablex Publishing: Norwood, 1984). 6 Providing consultative advice on diagnosis and antibiotic therapy for infectious diseases; q.v. BG Buchanan & EH Shortcliffe, Rule-Based Expert Systems: The MYCIN Experiments of the Stanford Heuristic Programming Project (Addison-Wesley: Reading, 1984). 7 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.11. 8 Ibidem p.15. 9 L Loevinger, "Jurimetrics: The Next Step Forward" (1949) Minnesota Law Review 33. 10 L Mehl, "Automation in the Legal World: From the Machine Processing of Legal Information to the 'Law Machine'" in Mechanisation of Thought Processes (HMSO: London, 1958) p.755. 11 RW Morrison, "Market Realities of Rule-Based Software for Lawyers: Where the Rubber Meets the Road" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 33 at p.35. 12 c.f. RA Clarke, Knowledge-Based Expert Systems (Working paper: Department of Commerce, Australian National University, 1988) p.6. 13 Primarily RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987). 14 Ibidem p.44; emphasis added. 15 MJ Sergot, "The Representation of Law in Computer Programs", Chapter One in TJM Bench-Capon, Knowledge-Based Systems and Legal Applications (Academic Press: London, 1991) at p.4. 16 P Harmon & D King, Expert Systems: Artificial Intelligence in Business (John Wiley & Sons: New York, 1985) at p.5. 17 R Forsyth, "The Anatomy of Expert Systems" Chapter Eight in M Yazdani (ed.), Artificial Intelligence: Principles and Applications (Chapman & Hall: London, 1986) pp.186-187. 18 R Forsyth, "The Anatomy of Expert Systems" Chapter Eight in M Yazdani, Artificial Intelligence: Principles and Applications (Chapman and Hall: London, 1986) p.194. 19 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.46. 20 Ibidem p.47. 21 F Hayes-Roth, DA Waterman & DB Lenat Building Expert Systems (Addison-Wesley: London, 1983) p.4. 22 e.g. The Latent Damage Adviser; q.v. PN Capper & RE Susskind, Latent Damage Law - The Expert System (Butterworths: London, 1988). 23 J Vaux, "AI and Philosophy: Recreating Naive Epistemology" Chapter Seven in KS Gill (ed.), Artificial Intelligence for Society (John Wiley & Sons: London, 1986) p.76. 24 T Winograd & F Flores, Understanding Computers and Cognition: A New Foundation for Design (Ablex: Norwood, 1986). 25 HL Dreyfus & SE Dreyfus, Mind over Machine (Basil Blackwell: Oxford, 1986). 26 A Hart & DC Berry, "Expert Systems in Perspective" in DC Berry & A Hart (eds) Expert Systems: Human Issues (MIT: Cambridge, 1990) p.11. 27 e.g. J Weizenbaum, Computer Power and Human Reason: From Judgment to Calculation (WH Freeman & Co: San Francisco, 1976). 28 R Hellawell, "A Computer Program for Legal Planning and Analysis: Taxation of Stock Redemptions" (1980) 80 Columbia Law Review 1363. See also NJ Bellord, "Tax Planning by Computer" in B Niblett (ed.), Computer Science and Law (Cambridge University Press: New York, 1980) p.173. 29 R Hellawell, "CHOOSE: A Computer Program for Legal Planning and Analysis" (1981) 19 Columbia Journal of Transnational Law 339. 30 R Hellawell, "SEARCH: A Computer Program for Legal Problem Solving" (1982) 15 Akron Law Review 635. 31 P Jackson, H Reichgelt & Fv Harmelen, Logic-Based Knowledge Representation (MIT: Cambridge, 1989). 32 Implemented in QUINTUS PROLOG; q.v. NH Minsky & D Rozenshtein, "System = Program + Users + Law" (1987) Proceedings First International Conference on Artificial Intelligence and Law 170. 33 Symbolic logic has had a profound influence in the artificial intelligence field; for a description see I Copi, Symbolic Logic (Macmillan: New York, 1973). 34 MJ Sergot, "A Brief Introduction to Logic Programming and Its Applications in Law" Chapter Five in C Walter (ed.) , Computer Power and Legal Language (Quorum: London, 1988) at pp.25-27. 35 C Mellish, "Logic Programming and Expert Systems" Chapter Nineteen in KS Gill (ed.), Artificial Intelligence for Society (John Wiley & Sons: London, 1986) at p.211. 36 F Hayes-Roth, DA Waterman & DB Lenat, Building Expert Systems (Addison-Wesley: London, 1983) at p.66. 37 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.208. 38 R Forsyth, "The Anatomy of Expert Systems" Chapter Eight in M Yazdani, Artificial Intelligence: Principles and Applications (Chapman and Hall: 1986) p.191. 39 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) pp.209-210. 40 RI Levine, DE Drang & B Edelson, Artificial Intelligence and Expert Systems (McGraw-Hill: 1990) Chapter Six, particularly at pp.62-65. 41 e.g. AW Koers & D Kracht, "A Goal Driven Knowledge Based System for a Domain of Private International Law" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 81. 42 q.v. RA Kowalski, "The Treatment of Negation in Logic Programs for Representing Legislation" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 11; P Asirelli, M De Santis & M Martelli, "Integrity Constraints in Logic Databases" (1985) 2 Journal of Logic Programming 221; K Eshghi & RA Kowalski, "Abduction Compared with Negation by Failure" (1989) Proceedings of the Sixth International Logic Programming Conference; and JW Lloyd, EA Sonenberg and RW Topot, "Integrity Constraint Checking in Stratified Databases" (1986) 4 Journal of Logic Programming 331. 43 TJM Bench-Capon, "Representating Counterfactual Conditionals" (1989) Proceedings Artificial Intelligence and the Simulation of Behvaiour 51. 44 "Augmented Prolog Expert System"; q.v. MJ Sergot, "A Brief Introduction to Logic Programming and Its Applications in Law" Chapter Five in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988) at pp.34-35. 45 P Hammond & MJ Sergot, "A PROLOG Shell for Logic Based Expert Systems" (1983) 3 Proceedings British Computer Society Expert Systems Conference. 46 DA Schlobohm, "A PROLOG Program Which Analyses Income Tax Issues under Section 318(a) of the Internal Revenue Code" in C Walter (ed.), Computing Power and Legal Reasoning (West Publishing: St Paul, 1985) p.765. 47 q.v. F Haft, RP Jones & T Wetter, "A Natural Language Based Legal Expert System for Consultation and Tutoring - The LEX Project" (1987) Proceedings First International Conference on Artificial Intelligence and the Law 75. 48 C Walter, "Elements of Legal Language" Chapter Three in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988). 49 LA Zadeh, "Fuzzy Sets" (1965) 8 Information and Control 338. 50 E Rosch & C Mervis, "Family Resemblances: Studies in the Internal Structure of Categories" (1975) 7 Cognitive Psychology 573. 51 M Novakowska, "Fuzzy Concepts: Their Strcuture and Problems of Measurement" in MM Gupta, RK Ragade & RR Yager (eds), Advances in Fuzzy Set Theory and Applications (North-Holland: Amsterdam, 1979) at p.361. 52 TJM Bench-Capon & MJ Sergot, "Toward a Rule-Based Representation of Open Texture in Law" Chapter Six in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988) at p.49. 53 D Berman & C Walter (ed.), "Toward a Model of Legal Argumentation" Chapter Four in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988) at p.22. 54 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.225. 55 CE Alchourrsn & AA Martino, "A Sketch of Logic Without Truth" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 165 at p.166. 56 HLA Hart, "Problems of the Philosophy of the Law" in HLA Hart, Essays in Jurisprudence and Philosophy (Clarendon Press: Oxford, 1983) p.100; and H Kelsen, "Law and Logic" in H Kelsen, Essays in Legal and Moral Philosophy (Reidel: Dordrecht, 1973) at p.229. 57 LT McCarty, "Permissions and Obligations - an Informal Introduction" (1983) Proceedings International Joint Conference on Artificial Intelligence-83; LT McCarty, "Permissions and Obligations - An Informal Introduction" in AA Martino & NF Socci (eds) Automated Analysis of Legal Texts (North-Holland: Amsterdam, 1986). Fora more developed system on the same principles, see H-N Castaqeda, "The Basic Logic for the Interpretation of Legal Texts" in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988) at p.167. 58 GHv Wright, "Deontic Logic" (1951) 60 Mind 1. 59 McCarty suggests that it is helpful to think of the set as an "oracle" to be consulted when contemplating a course of action; see LT McCarty, "Permissions and Obligations - an Informal Introduction" in AA Martino & NF Socci (eds) Automated Analysis of Legal Texts (North-Holland: Amsterdam, 1986). 60 Note LT McCarty, "Permissions and Obligations - A Informal Introduction" in AA Martino & NF Socci (eds) Automated Analysis of Legal Texts (North-Holland: Amsterdam, 1986)Definitions 5-7. 61 LT McCarty, "Clausal Intuitionistic Logic I: Fixed-Point Semantics" (1988) 5 Journal of Logic Programming 1; LT McCarty, "Clausal Intuitionistic Logic II: Tableau Proof Procedures" (1988) 5 Journal of Logic Programming 93. 62 LT McCarty, "On the Role of Prototypes in Appellate Legal Argument" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 185 at p.187. 63 AJI Jones, "On the Relationship Between Permission and Obligation" (1987) Proceedings First International Conference on Artificial Intelligence and Law 164 at pp.166-168. 64 LT McCarty & Rvd Meyden, "Indefinite Reasoning with Definite Rules" (1991) Proceedings of the Twelfth International Joint Conference on Artificial Intelligence. 65 Particularly those of consequential closure; AJI Jones & I Pren, "Ideality, Sub-Ideality and Deontic Logic" (1985) 2 Synthise 65. 66 R Stamper, "The LEGOL-1 Prototype System and Language" (1977) 20 The Computer Journal 102. 67 R Stamper, C Tagg, P Mason, S Cook & J Marks, "Developing the LEGOL Semantic Grammar" in C Ciampi (ed.) Artificial Intelligence and Legal Information Systems (North-Holland: Amsterdam, 1982) p.357. 68 R Stamper, "LEGOL: Modelling Legal Rules by Computer" in B Niblett (ed.), Computer Science and Law (Cambridge University Press: New York, 1980) p.45. 69 R Stamper, "A Non-Classical Logic for Law Based on the Structures of Behaviour" in AA Martino & F Socci (eds), Automated Analysis of Legal Texts (North-Holland: Amsterdam, 1986) p.57. 70 S Jones, "Control Structures in Legislation" in B Niblett (ed.), Computer Science and Law (Cambridge University Press: New York, 1980) p.157. 71 MJ Sergot, Programming Law: LEGOL as a Logic Programming Language (Imperial College: London, 1980). 72 MJ Sergot, "The Representation of Law in Computer Programs", Chapter One in TJM Bench-Capon, Knowledge-Based Systems and Legal Applications (Academic Press: London, 1991) at p.35. 73 IE Pratt, Epistemology and Artificial Intelligence (PhD dissertation: Princeton, 1987) p.18; emphasis in original. 74 Susskind and Gold. 75 Including Bench-Capon, Cordingley, Forder, Frohlich, Gilbert, Luff, Protman, Sergot, Storrs and Taylor; q.v. RN Moles, "Logic Programming - An Assessment of Its Potential for Artificial Intelligence Applications in Law" (1991) 2 Journal of Law and Information Science 137 at pp.146-147. 76 RE Susskind, "The Latent Damage System" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 23 at p.29. 77 On causality, note CG de'Bessonet & CR Cross, "Representation of Some Aspects of Causality" in C Walter (ed.) Computing Power and Legal Reasoning (West: St Paul, 1985) pp.205-214. 78 e.g. SR Goldman, MG Dyer & M Flowers, "Precedent-based Legal Reasoning and Knowledge Acquisition in Contract Law: a Process Model" (1987) Proceedings First International Conference on Artificial Intelligence and Law 210 at pp.214-215 using Hohfeldian analysis of rights; q.v. WN Hohfeld, "Some Fundamental Legal Conceptions As Applied in Judicial Reasoning" (1917) 23 Yale Law Journal 16. 79 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.227. 80 P Leith, "Logic, Formal Models and Legal Reasoning" (1984) Jurimetrics Journal 334 at pp.335-336. 81 N MacCormick, Legal Reasoning and Legal Theory (Oxford University Press: Oxford, 1978) at p.37. 82 MJ Sergot, "A Brief Introduction to Logic Programming and Its Applications in Law" Chapter Five in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988) at p.26. 83 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) pp.181-198, particularly at p.188. 84 LE Allen & CS Saxon, "Some Problems in Designing Expert Systems to Aid Legal Reasoning" (1987) Proceedings First International Conference on Artificial Intelligence and Law 94 at p.94. 85 RE Susskind, "The Latent Damage System" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 23 at p.28. 86 Ibidem p.30. 87 AvdL Gardner, "Overview of an AI Approach to Legal Reasoning" in C Walter (ed.),Computing Power and Legal Reasoning (West: St Paul, 1985) p.247. 88 HLA Hart, "Positivism and the Separation of Law and Morals" (1958) 79 Harvard Law Review 593 at p.599. 89 G Gottlieb, The Logic of Choice: An Investigation of the Concepts of Rule and Rationality (Allen & Unwin: London, 1968) p.17. 90 OC Jensen, The Nature of Legal Argument (Basil Blackwell: Oxford, 1957) p.16; A Wilson, "The Nature of Legal Reasoning: A Commentary with Special Reference to Professor MacCormick's Theory" (1982) 2 Legal Studies 269 at pp.278-280. 91 MJ Detmold, The Unity of Law and Morality: A Refutation of Legal Positivism (Routledge & Kegan Paul: London, 1984) p.15; c.f. RE Susskind, "Detmold's Refutation of Positivism and the Computer Judge" (1986) 49 Modern Law Review 125. 92 HLA Hart, "Positivism and the Separation of Law and Morals" (1958) 79 Harvard Law Review 593 at p.607. 93 DB Skalak, "Taking Advantage of Models for Legal Classification" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 234. 94 CD Hafner, "Conceptual Organisation of Case Law Knowledge Bases" (1987) Proceedings First International Conference on Artificial Intelligence and Law 35 at pp.36-37. 95 Note the diagram attached to RE Susskind, "The Latent Damage System" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 23. 96 MJ Sergot, "Representing Legislation as Logic Programs" (1985) 11 Machine Intelligence 209. 97 MJ Sergot, F Sadri, RA Kowalski, F Kriwaczek, P Hammond, & HT Cory, "The British Nationality Act as a Logic Program" (1986) 29 Communications of the ACM 370; and MJ Sergot, HT Cory, P Hammond, RA Kowalski, F Kriwaczek, & F Sadri, "Formalisation of the British Nationality Act" in C Arnold (ed.), Yearbook of Law, Computers and Technology (Butterworths: London, 1986). 98 TJM Bench-Capon, GO Robinson, TW Routen & MJ Sergot, "Logic Programming for Large Scale Applications in Law: A Formalisation of Supplementary Benefit Legislation" (1987) Proceedings First International Conference on Artificial Intelligence and Law 190. 99 q.v. P Johnson & D Mead, "Legislative Knowledge Base Systems for Public Administration: Some Practical Issues" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 74. 100 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.100. 101 The concept can be attributed to Quillian; q.v. MR Quillian, "Word Concepts: A Theory and Simulation of Some Basic Semantic Capabilities" (1967) 12 Behvioural Science 410. 102 WA Woods, "What's in a Link: Foundations for Semantic Networks" in DG Bobrow & AM Collins (eds), Representation and Understanding: Studies in Cognitive Science (Academic Press: New York, 1975) p.32. 103 M Minsky, "A Framework for Representing Knowledge" in J Haugeland (ed.), Mind Design (MIT Press: Cambridge, 1981) p.95. 104 PJ Hayes, "The Logic of Frames" in D Metzing (ed.), Frame Conceptions and Text Understanding (Walter de Gruyter: Berlin, 1979) p.46. 105 MJ Sergot, "The Representation of Law in Computer Programs", Chapter One in TJM Bench-Capon, Knowledge-Based Systems and Legal Applications (Academic Press: London, 1991) at p.48. 106 LT McCarty, "Reflections on TAXMAN: An Experiment in Artificial Intelligence and Legal Reasoning" (1977) 90 Harvard Law Review 837; and LT McCarty, "The TAXMAN Project: Towards a Cognitive Theory of Legal Argument" in B Niblett (ed.), Computer Science and Law (Cambridge University Press: New York, 1980). 107 PJ Hayes, "The Logic of Frames" in D Metzing (ed.), Frame Conceptions and Text Understanding (de Gruyter: New York, 1979). 108 Example adapted from MJ Sergot, "The Representation of Law in Computer Programs", Chapter One in TJM Bench-Capon, Knowledge-Based Systems and Legal Applications (Academic Press: London, 1991) at p.46. 109 KE Sanders, "Representing and Reasoning About Open-Textured Predicates" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 137 at p.138. 110 LT McCarty & NS Sridharan, "The Representation of an Evolving System of Legal Concepts II: Prototypes and Deformations" (1987) Proceedings of the Seventh International Joint Conference on Artificial Intelligence 246. 111 TJM Bench-Capon & MJ Sergot, "Toward a Rule-Based Representation of Open Texture in Law" Chapter Six in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988) at p.47. 112 SS Weiner, "Reasoning About "Hard" Cases in Talmudic Law" (1987) Proceedings First International Conference on Artificial Intelligence and Law 222 at p.223. 113 P Leith, "Logic, Formal Models and Legal Reasoning" (1984) Jurimetrics Journal p.334 at p.356. 114 KA Lambert & MH Grunewald, "LESTER: Using Paradigm Cases in a Quasi-Prcedential Legal Domain" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 87. 115 J Popple, "Legal Expert Systems: The Inadequacy of a Rule-based Approach" (1991) 23 Australian Computer Journal 11 at p.15. 116 Also note the GREBE system; q.v. LK Branting, "Representing and Reusing Explanations of Legal Precedents" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 103. 117 RA Kowalski, "Case-based Reasoning and the Deep Structure Approach to Knowledge Representation" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 21 at p.23. 118 KD Ashley & EL Rissland, "Waiting on Weighting: a Symbolic Least Commitment Approach" (1988) Proceedings American Association for Artificial Intelligence. 119 MJ Sergot, "The Representation of Law in Computer Programs", Chapter One in TJM Bench-Capon, Knowledge-Based Systems and Legal Applications (Academic Press: London, 1991) at p.65. 120 J Zeleznikow, Building Intelligent Legal Tools - The IKBALS Project (1991) 2 Journal of Law and Information Science 165 at p.173. 121 EL Rissland & KD Ashley, "A Case-Based System for Trade Secrets Law" (1987) Proceedings First International Conference on Artificial Intelligence and Law 60. 122 M Betzer, "Legal Resoning in 3-D" (1987) Proceedings First International Conference on Artificial Intelligence and Law 155 at p.155. 123 RA Kowalski, "Case-based Reasoning and the Deep Structure Approach to Knowledge Representation" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 21. 124 RA Kowalski, "Case-based Reasoning and the Deep Structure Approach to Knowledge Representation" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 21 at p.26. 125 G Greenleaf, A Mowbray & AL Tyree, "Expert Systems in Law: The DATALEX Project" (1987) Proceedings First International Conference on Artificial Intelligence and Law 9 at p.12. 126 e.g. SR Goldman, MG Dyer & M Flowers, "Precedent-based Legal Reasoning and Knowledge Acquisition in Contract Law: a Process Model" (1987) Proceedings First International Conference on Artificial Intelligence and Law 210; and MT MacCrimmon, "Expert Systems in Case-Based Law: The Hearsay Rule Adviser" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 68. 127 G Vossoss, J Zeleznikow & T Dillon, "Combining Analogical and Deductive Reasoning in Legal Knowledge Base Systems - IKBALS II" in Cv Noortwijk, AHJ Schmidt & RGF Winkels (eds), Legal Knowledge Based Systems: Aims for Research and Development (Koninklijke: Lelystad, 1991) p.97 at p.100. 128 The weighting scheme used by Kowalski was: Highest level court = 70; appeal level court = 50; trial level court = 30. Add 10 points for trial or appeals local to the jurisdiction. Deduct 15 points for foreign jurisdictions, except England, then 10 points. Add 1 to 5 points if case is recent: 1986 = 1 to 1990 = 5. See RA Kowalski, Case-based Reasoning and the Deep Structure Approach to Knowledge Representation (1991) Proceedings Third International Conference on Artificial Intelligence and Law 21. 129 G Vossos, T Dillon & J Zeleznikow, "The Use of Object Oriented Principles to Develop Intelligent Legal Reasoning Systems" (1991) 23 Australian Computer Journal 2. 130 J Zeleznikow & D Hunter, "Rationales for the Continued Development of Legal Expert Systems" (1992) 3 Journal of Law and Information Science 94 at pp.102-103. 131 RE Susskind, "Expert Systems in Law: A Jurisprudential approach to Artificial Intelligence and Legal Reasoning" (1986) 49 Modern Law Review 168 at p.171; see also RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.20. 132 RA Kowalski, "Case-Based Reasoning and the Deep Structure Approach to Knowledge Representation" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 21 at p.21. 133 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) pp.81-82. 134 Ibidem pp.21-23. 135 TJM Bench-Capon & J Forder, "Knowledge Representation for Legal Applications" Chapter Twelve in TJM Bench-Capon, Knowledge-Based Systems and Legal Applications (Academic Press: London, 1991) at p.249. 136 HJ Levesque & RJ Brachman, "A Fundamental Tradeoff in Knowledge Representation and Reasoning" Chapter Four in RJ Brachman & HJ Levesque (eds), Readings in Knowledge Representation (Morgan Kaufmann: Los Altos, 1985) at pp.66-67. 137 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.49. 138 q.v. TJM Bench-Capon & F Coenen, "Exploiting Isomorphism: Development of a KBS to Support British Coal Insurance Claims" (1991) Proceedings Third International Conference of Artificial Intelligence and Law 62. 139 RN Moles, "Logic Programming - An Assessment of Its Potential for Artificial Intelligence Applications in Law" (1991) 2 Journal of Law and Information Science 137 at p.144. 140 Bench-Capon's own words; q.v. TJM Bench-Capon & J Forder, "Knowledge Representation for Legal Applications" Chapter Twelve in TJM Bench-Capon, Knowledge-Based Systems and Legal Applications (Academic Press: London, 1991) at p.259. 141 e.g. C Biagioli, P Mariani & D Tiscornia, "ESPLEX: a Rule and Conceptual Based Model for Representing Statutes" (1987) Proceedings First International Conference on Artificial Intelligence and Law 240, and the examples previously cited. 142 e.g. DM Sherman, "A Prolog Model of the Income Tax Act of Canada" (1987) Proceedings First International Conference on Artificial Intelligence and Law 127; also note TAXMAN and like projects cited. 143 B Niblett, "Computer Science and Law: An Introductory Discussion" in B Niblett (ed.), Computer Science and Law (Cambridge University Press: Cambridge, 1980) at pp.16-17. 144 For instance into ANF (Atomically Normalised Form) used with the CCLIPS system (Civil Code Legal Information Processing System); q.v. G Cross, CGde Bossonet, T Bradshaw, G Durham, R Gupta & M Nasiruddin, "The Implementation of CCLIPS" Chapter Nine in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988) p.90. 145 JP Dick, "Conceptual Retrieval and Case Law" (1987) Proceedings First International Conference on Artificial Intelligence and Law 106 at p.109; although such material may classify as a source of heuristics. Susskind does not address this, however. 146 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) pp.84-85; citing HLA Hart, The Concept of Law (Clarendon Press: Oxford, 1961) p.131. 147 KE Sanders, "Representing and Reasoning About Open-Textured Predicates" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 137 at p.142. 148 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.61; contra G Greenleaf, A Mowbray & AL Tyree, "Expert Systems in Law: The DATALEX Project" (1987) Proceedings First International Conference on Artificial Intelligence and Law 9. 149 c.f. LB Solum, "On the Indeterminacy Crisis: Critiquing Critical Dogma" (1987) 54 University of Chicago Law Review 462. 150 J Boyle, "Anatomy of a Torts Class" (1985) 34 American University Law Review 131; see also M Kelman, "Trashing" (1984) 36 Stanford Law Review 293. 151 A Mason, "Future Directions in Australian Law" (1987) 13 Monash Law Review 149, particularly at pp.154-155 and p.158; FG Brennan, "Judicial Method and Public Law" (1979) 6 Monash Law Review 12; and M McHugh, "The Law-making Function of the Judicial Process" (1988) 62 Australian Law Journal 15. 152 e.g. JC Smith and C Deedman, "The Application of Expert Systems Technology to Case-Based Reasoning" (1987) Proceedings First International Conference on Artificial Intelligence and Law 84. 153 E Levi, An Introduction to Legal Reasoning (University of Chicago Press: Chicago, 1949) pp.3-5. 154 AL Goodhart, "The Ratio Decidendi of a Case" (1930) 40 Yale Law Journal 161. 155 J Stone, "The Ratio of the Ratio Decidendi" in Lord Lloyd & MDA Freeman, Lloyd's Introduction to Jurisprudence (5th Ed.) (Stevens: London, 1985) p.1164. 156 "It is unclear". 157 S Strvmholm, Rdtt. rottskollor och rottssystem (3rd Ed.) (Norstedts: Stockholm, 1987) cited by P Wahlgren, "Legal Reasoning - A Jurisprudence Description" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 147 at p.148. 158 TJM Bench-Capon & F Coenen, "Practical Application of KBS to Law: The Crucial Role of Maintenance" in Cv Noortwijk, AHJ Schmidt & RGF Winkels (eds), Legal Knowledge Based Systems: Aims for Research and Development (Koninklijke: Lelystad, 1991) p.5. 159 P Bratley, J Frimont, E Mackaay & D Poulin, "Coping with Change" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 69. 160 e.g. P Hammond, "Representation of DHSS Regulations as a Logic Program" (1983) Proceedings of the 3rd British Computer Society Expert Systems Conference 225; and the Estate Planning System; q.v. DA Schlobohm & DA Waterman, "Explanation for an Expert System that Performs Estate Planning" (1987) Proceedings First International Conference on Artificial Intelligence and Law 18. 161 RA Kowalski & MJ Sergot, "The Use of Logical Models in Legal Problem Solving" (1990) 3 Ratio Juris 201 at p.207. 162 DA Schlobohm & LT McCarty, "EPS II: Estate Planning With Prototypes" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 1. 163 P Bratley, J Frimont, E Mackaay & D Poulin, "Coping with Change" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 69. 164 AvdL Gardner, "Representing Developiong Legal Doctrine" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 16 at p.21. 165 Ibidem p.19. 166 A narrow definition of "information" is a common criticism of modern expert systems; q.v. HL Dreyfus & SE Dreyfus, Mind over Machine (Basil Blackwell: Oxford, 1986); T Roszak, "The Cult of Information" (Pantheon: London, 1986); DR Hofstadter, Metamagical Themas (Penguin Press: London, 1985); and T Winograd & F Flores, Understanding Computers and Cognition: A New Foundation for Design (Ablex: Norwood, 1986). 167 RN Moles, "Logic Programming - An Assessment of Its Potential for Artificial Intelligence Applications in Law" (1991) 2 Journal of Law and Information Science 137 at p.144. 168 C Biagioli, P Mariani & D Tiscornia, "ESPLEX: a Rule and Conceptual Based Model for Representing Statutes" (1987) Proceedings First International Conference on Artificial Intelligence and Law 240 at p.241. 169 C Smith & C Deedman, "The Application of Expert Systems Technology to Case-Based Reasoning" (1987) Proceedings First International Conference on Artificial Intelligence and Law 84 at p.87. 170 RE Susskind, "Expert Systems in Law - Out of the Research Laboratory and into the Marketplace" (1987) Proceedings First International Conference on Artificial Intelligence and Law 1 at p.2. 171 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.78. 172 JC Smith & C Deedman, "The Application of Expert Systems Technology to Case-Based Reasoning" (1987) Proceedings First International Conference on Artificial Intelligence and Law 84 at p.85. 173 Approximately 50 texts and 100 articles; q.v. RE Susskind, "Expert Systems in Law - Out of the Research Laboratory and into the Marketplace" (1987) Proceedings First International Conference on Artificial Intelligence and Law 1 at p.2; note the similarity to the issue of epistemology of law. 174 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.27. 175 For instance, Dworkin's "principles"; q.v. RM Dworkin, Taking Rights Seriously (Duckworth: London, 1977), RM Dworkin, A Matter of Principle (Harvard University Press: London, 1985), and RM Dworkin, Law's Empire (Fontana: London, 1986). 176 RE Susskind, "Expert Systems in Law - Out of the Research Laboratory and into the Marketplace" (1987) Proceedings First International Conference on Artificial Intelligence and Law 1 at p.3. 177 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.254. 178 B Niblett, "Expert Systems for Lawyers" (1981) 29 Computers and Law 2 at p.3. 179 PJ Hayes, "On the Differences Between Psychology and Artificial Intelligence" in M Yazdani & A Narayanan, Artificial Intelligence: Human Effects (Ellis Horwood: London, 1984) p.158. 180 DR Hofstader, Gvdel, Escher, Bach: An Eternal Golden Braid (Harvester Press: New York, 1979) p.578. 181 RA Kowalski, "Leading Law Students to Uncharted Waters and Making them Think: Teaching Artificial Intelligence and Law" (1991) 2 Journal of Law and Information Science 185 at p.187 nt.5. 182 D Brown, "The Third International Conference on Artificial Intelligence and Law: Report and Comments" (1991) 2 Journal of Law and Information Science 233 at p.238. 183 B Niblett, "Expert Systems for Lawyers" (1981) 29 Computers and Law 2 at p.3. 184 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.7. 185 P Leith, "Clear Rules and Legal Expert Systems" in AA Martino & F Socci (eds), Automated Analysis of Legal Texts (North-Holland: Amsterdam, 1986) p.661; and P Leith, "Fundamental Errors in Legal Logic Programming" (1986) 3 The Computer Journal 29. 186 LT McCarty, "Some Requirements for a Computer-based Legal Consultant" (Research Report: Rutgers University, 1980) at pp.2-3, cited in RN Moles, Definition and Rule in Legal Theory: A Reassessment of HLA Hart and the Positivist Tradition (Basil Blackwell: Oxford, 1987) p.269; emphasis added. 187 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.53. 188 MA Boden, Artificial Intelligence and Natural Man (Basic Books: New York, 1977), cited in D Partridge, "Social Implications of Artificial Intelligence" Chapter Thirteen in M Yazdani (ed.), Artificial Intelligence: Principles and Applications (Chapman & Hall: London, 1986) at p.326. See also D Partridge, Artificial Intelligence: Applications in the Future of Software Engineering (Ellis Horwood, Chichester, 1986). 189 RN Moles, "Logic Programming: An Assessment of its Potential for Artificial Intelligence Applications in Law" (1991) 2 Journal of Law and Information Science 137 at p.161. 190 TJM Bench-Capon, "Deep Models, Normative Reasoning and Legal Expert Systems" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 37 at p.37. 191 Glauoma diagnosis system. 192 LT McCarty, "Intelligent Legal Information Systems: Problems and Prospects" in CM Campbell (ed.), Data Processing and the Law (Sweet & Maxwell: London, 1984) p.126. 193 SS Weiner, "Reasoning About "Hard" Cases in Talmudic Law" (1987) Proceedings First International Conference on Artificial Intelligence and Law 222 at p.223. 194 MJ Sergot, HT Cory, P Hammond, RA Kowalski, F Kriwacek & F Sadri, "Formalisation of the British Nationality Act" (1986) 2 Yearbook of Law Computers and Technology; and TJM Bench-Capon, GO Robinson, TW Routen & MJ Sergot, "Logic Programming for Large Scale Applications in Law" (1987) Proceedings First International Conference on Artificial Intelligence and Law 190. 195 JC Smith & C Deedman, "The Application of Expert Systems Technology to Case-Based Reasoning" (1987) Proceedings First International Conference on Artificial Intelligence and Law 84 at p.85. 196 RA Kowalski, Case-based Reasoning and the Deep Structure Approach to Knowledge Representation (1991) Proceedings Third International Conference on Artificial Intelligence and Law 21 at p.22. 197 LT McCarty & NS Sridharan, "The Representation of an Evolving System of Legal Concepts II: Prototypes and Deformations" (1987) Proceedings of the Seventh International Joint Conference on Artificial Intelligence 246 at p.250. 198 D Makinson, "How to Give it Up: A Survey of Some Formal Aspects of the Logic of Theory Change" (1985) 62 Synthise 347. 199 P Bratley, J Frimont, E Mackaay & D Poulin, "Coping with Change" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 69 at p.74. 200 See RM Dworkin, Law's Empire (Fontana: London, 1986) pp.250-254 on the difficulty in "compartmentalization" of the law. 201 LT McCarty, "Some Requirements for a Computer-based Legal Consultant" (Research Report: Rutgers University, 1980) cited in MJ Sergot, "The Representation of Law in Computer Programs", Chapter One in TJM Bench-Capon, Knowledge-Based Systems and Legal Applications (Academic Press: London, 1991) at pp.46-47. 202 P Leith, "Logic, Formal Models and Legal Reasoning" (1984) Jurimetrics Journal p.334 at p.356. 203 NE Simmonds, "Between Positivism and Idealism" (1991) 50 Cambridge Law Journal 308 at pp.312-313. 204 M Minsky, "A Framework for Representing Knowledge" in J Haugeland (ed.), Mind Design (MIT Press: Cambridge, 1981) p.95 at p.100. 205 It is also ironic in light of Hart's alleged reliance on Wittgenstein's linguistic philosophy; q.v. Cotterrell, The Politics of Jurisprudence: A Critical Introduction to Legal Philosophy (Butterworths: London, 1989) pp.89-90. 206 J Vaux, "AI and Philosophy: Recreating Naive Epistemology" Chapter Seven in KS Gill (ed.), Artificial Intelligence for Society (John Wiley & Sons: London, 1986) p.76.; q.v. L Wittgenstein, Philosophical Investigations (Basil Blackwell: London, 1953). 207 RA Kowalski & MJ Sergot, "The Uses of Logical Models in Legal Problem Solving" (1990) 3 Ratio Juris 201 at p.205. 208 See HLA Hart, "Definition and Theory in Jurisprudence" (1954) 70 Law Quarterly Review 37. 209 L Fuller, "Positivism and Fidelity to Law - A Reply to Professor Hart" (1958) 71 Harvard Law Review 630 at p.666. 210 Raz's approach to adjudication shares similar characteristics; q.v. J Raz, "The Problem about the Nature of Law" (1983) 31 University of Western Ontatio Law Review 202 at pp.213-216. 211 HLA Hart, The Concept of Law (Clarendon Press: Oxford, 1961) p.126. 212 KR Popper, Conjectures and Refutations (4th Ed.) (Routledge and Kegan Paul: London, 1972) p.46. 213 JW Harris, Law and Legal Science (Clarendon Press: Oxford, 1979) p.166. 214 M Polyani, Personal Knowledge - Towards a Post-Critical Philosophy (Routledge and Kegan Paul: London, 1958). 215 DC Berry, "The Problem of Implicit Knowledge" (1987) 4 Expert Systems 144. 216 DC Berry & A Hart, "The Way Forward" in DC Berry & A Hart (eds) Expert Systems: Human Issues (MIT: Cambridge, 1990) p.256. 217 E Husserl, Cartesian Meditations (Martinus Nijhoff: The Hague, 1960) pp.54-55. 218 B MacLennan, "Logic for the New AI" in JH Fetzer (ed.), Aspects of Artificial Intelligence (Kluwer: Dordrecht, 1988) at p.163. 219 C Hempel, Philosophy of Natural Science (Prentice Hall: London, 1966). 220 A Narayanan, "Why AI Cannot be Wrong" Chapter Five in KS Gill (ed.), Artificial Intelligence for Society (John Wiley & Sons: London, 1986) at p.48. 221 P Davies, "Living in a Non-Maaterial World - the New Scientific Consciousness" (1991) The Australian (9th October) pp.18-19 at p.19. 222 RS Pound, "Mechanical Jurisprudence" (1908) 8 Columbia Law Review 605. 223 J Searle, "Minds, Brains and Programs" (1980) 3 Behavioural and Brain Sciences 417. 224 RE Susskind, Expert Systems in Law: A Jurisprudential Inquiry (Clarendon Press: Oxford, 1987) p.241. 225 RHS Tur, "Positivism, Principles, and Rules" in E Atwool (ed.), Perspectives in Jurisprudence (University of Glascow Press: Glascow, 1997) at p.51. 226 EL Rissland & DB Skalak, "Interpreting Statutory Predicates" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 46 at p.46. 227 MJ Detmold, "Law as Practical Reason" (1989) 48 Cambridge Law Journal 436 at p.460. 228 Ibidem p.439. 229 L Fuller, "Positivism and Fidelity to Law - A Reply to Professor Hart" (1958) 71 Harvard Law Review 630 at p.663; see also RS Summers, "Professor Fuller on Morality and Law" in RS Summers (ed.), More Essays on Legal Philosophy: General Assessment of Legal Philosophies (Basil Blackwell: Oxford, 1971) at pp.117-119. 230 F Schauer, Playing by the Rules: A Philosophical Examination of Rule-based Decision-Making in Law and in Life (Clarendon Press: Oxford, 1991) at pp.59-60. 231 HLA Hart, The Concept of Law (Clarendon Press: Oxford, 1961) p.56. 232 H Williamson, "Some Implications of Acceptance of Law as Rule Structure" (1967) 3 Adelaide Law Review 18 at pp.42-43. 233 c.f. A Glass, "Interpretive Practices in Law and Literary Criticism" (1991) 7 Australian Journal of Law & Society 16. 234 DN Herman, "Phenomonology, Structuralism, Hermeneutics, and Legal Study: Applications of Contemporary Continental Though to Legal Phenomena" (1982) 36 University of Miami Law Review 379. 235 P Linzer, "Precise meaning and Open Texture in Legal Writing and Reading" Chapter Two in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988). 236 M Weait, "Swans Reflecting Elephants: Imagery and the Law" (1992) 3 Law and Critique 59 at p.66. 237 P Gabel & P Harris, "Building Power and breaking Images: Critical Legal Theory and the Practice of Law" (1982-83) 11 Review of Law & Social Change 369 at p.370. 238 HL Dreyfus, "From Micro-Worlds to Knowledge Representation: AI at an Impasse" in J Haugeland (ed.), Mind Design (MIT Press: Cambridge, 1981) p.161 at p.170. 239 DG Bobrow & T Winograd, "An Overview of KRL, A Knowledge Representation Language" (1977) 1 Cognitive Science 3 at p.32. 240 C Fried, "Sonnet LXV and the 'Black Ink' of the Framer's Intention" (1987) 100 Harvard Law Review 751 at pp.757-758. 241 J Weizenbaum, Computer Power and Human Reason: From Judgment to Calculation (WH Freeman & Co: San Francisco, 1976) cited in D Partridge, "Social Implications of Artificial Intelligence" Chapter Thirteen in M Yazdani (ed.), Artificial Intelligence: Principles and Applications (Chapman & Hall: London, 1986) at pp.330-331. 242 Contra note MA Boden, "AI and Human Freedom" in M Yazdani & A Narayanan (eds), Artificial Intelligence: Human Effects (Ellis Horwood: Chichester, 1984). 243 C Tapper, "Lawyers and Machines" (1963) 26 Modern Law Review 121. 244 Ibidem p.128. 245 c.f. TJM Bench-Capon, "Deep Models, Normative Reasoning and Legal Expert Systems" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 37 at p.42. 246 J Zeleznikow, "Building Intelligent Legal Tools - The IKBALS Project" (1991) 2 Journal of Law and Information Science 165. 247 e.g. DE Wolstenholme, "Amalgamating Regulation and Case-based Advice Systems through Suggested Answers" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 63. 248 c.f. R Wright, "The Cybernauts have Landed" (1991) Law Institute Journal 490 at p.491. 249 C Tapper, "Lawyers and Machines" (1963) 26 Modern Law Review 121 at p.126. 250 q.v. PJ Ward, "Computerisation of Legal Material in Australia" (1982) 1 Journal of Law and Information Science 162. 251 J Bing, "The Text Retrieval System as a Conversation Partner" in C Arnold (ed.) Yearbook of Law, Computers and Technology (Butterworths: London, 1986) p.25. 252 G Greenleaf, "Australian Approaches to Computerising Law - Innovation and Integration" (1991) 65 Australian Law Journal 677. 253 SJ Latham, "Beyond Boolean Logic: Probabilistic Approaches to Text Retrieval" (1991) 22 The Law Librarian 157. 254 J Bing, "Legal Text Retrieval Systems: The Unsatisfactory State of the Art" (1986) 2 Journal of Law and Information Science 1 at pp.16-17. 255 RM Tong, CA Reid, GJ Crowe & PR Douglas, "Conceptual Legal Document Retrieval Using the RUBRIC System" (1987) Proceedings First International Conference on Artificial Intelligence and Law 28; and J Bing, "Designing Text Retrieval Systems for 'Conceptual Searching'" (1987) Proceedings First International Conference on Artificial Intelligence and Law 43. 256 J Kolodner, "Maintaining Organisation in a Dynamic Long-Term Memory" (1983) 7 Cognitive Science; CD Hafner, "Conceptual Organisation of Case Law Knowledge Bases" (1987) Proceedings First International Conference on Artificial Intelligence and Law 35. 257 T Mitchell, "Learning and Problem Solving" (1983) Proceedings of International Joint Conference on Artificial Intelligence. 258 DE Rose & RK Belew, "Legal Information Retrieval: A Hybrid Approach" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 138. 259 CD Hafner, "Conceptual Organisation of Case Law Knowledge Bases" (1987) Proceedings First International Conference on Artificial Intelligence and Law 35. 260 LT McCarty, "On the Role of Prototypes in Appellate Legal Argument" (1991) Proceedings Third International Conference on Artificial Intelligence and Law 185 at p.186. 261 TJM Bench-Capon & MJ Sergot, "Toward a Rule-Based Representation of Open Texture in Law" Chapter Six in C Walter (ed.), Computer Power and Legal Language (Quorum: London, 1988) at p.58. 262 KD Ashley, "Toward a Computational Theory of Arguing with Precedents: Accomodating Multiple Interpretations of Cases" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 99; and KD Ashley & EL Rissland, "But See, Accord: Generating 'Blue Book' Citations in HYPO" (1987) Proceedings First International Conference on Artificial Intelligence and Law 67. 263 EL Rissland, "Examples in Legal Reasoning: Legal Hypotheticals" (1983) Proceedings Eighth International Joint Conference on Artificial Intelligence 90; EL Rissland & EM Soloway, "Overview of an Example Generation System" (1980) Proceedings First Annual National Conference on Artificial Intelligence; and EL Rissland, EM Valcarce & KD Ashley, "Explaining and Arguing with Examples" (1984) Proceedings National Conference on Artificial Intelligence. 264 CC Marshall, "Representing the Structure of a Legal Argument" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 121. On the structure of legal argument see S Toulmin, The Uses of Argument (Cambridge University Press: Cambridge, 1958); S Toulmin, RD Reike, & A Janik, An Introduction to Reasoning (MacMillan Press: New York, 1979); and C Perelman, The Idea of Justice and the Problem of Argument (Routledge & Kegan Paul: London, 1963). 265 KD Ashley & EL Rissland, "Toward Modelling Legal Argument" in AA Martino & F Socci (eds), Automated Analysis of Legal Texts (North-Holland: Amsterdam, 1986) at p.19; also KD Ashley, "Toward a Computational Theory of Arguing with Precedents" (1989) Proceedings Second International Conference on Artificial Intelligence and Law 93. 266 EL Rissland, "Learning How to Argue: Using Hypotheticals" (1984) Proceedings First Annual Conference on Theoretical Issues in Conceptual Information Processing; EL Rissland, "Argument Moves and Hypotheticals" in C Walter (ed.), Computing Power and Legal Reasoning (West Publishing: St Paul, 1985). 267 RH Michaelson, "An Expert System for Federal Tax Planning" (1984) 1 Expert Systems 2. 268 e.g. the Retirement Pension Forecast and Advice System (relying on the Aion Development System shell) q.v. S Springel-Sinclair & G Trevena, "The DHSS Retirement Pension Forecast and Advice System" in P Duffin (ed.) Knowledge Based Systems: Applications in Administrative Government (Ellis Horwood: Chichester, 1988). 269 G De Jong, "Towards a Model of Conceptual Knowledge Acquisition Through Directed Experimentation" (1983) Proceedings of International Joint Conference on Artificial Intelligence. 270 Legal Decision-making System; q.v. DA Waterman & MA Peterson, "Rule-based Models of Legal Expertise" (1980) Proceedings First Annual National Conference on Artificial Inelligence 272; and DA Waterman & MA Peterson, "Evaluating Civil Claims: An Expert Systems Approach" (1984) Expert Systems 1. 271 q.v. DA Waterman, RH Anderson, F Hayes-Roth, P Klahr, G Martins & SJ Rosenschein, Design of a Rule-Oriented System for Implementing Expertise (Rand Corporation: Santa Monica, 1979). 272 System for Asbestos Litigation; q.v. DA Waterman, J Paul & MA Peterson, "Expert Systems for Legal Decision Making" (1986) 4 Expert Systems 212. 273 G Greenleaf, "Australian Approaches to Computerising Law - Innovation and Integration" (1991) 65 Australian Law Journal 677 at p.679. 274 SS Nagel & R Barczyk, "Can Computers Aid the Dispute Resolution Process?" (1988) 71 Judicature 253. 275 q.v. WM Bain, Toward a Model of Subjective Interpretation (Department of Commerce Research Report: Yale University, 1984) cited in MJ Sergot, "The Representation of Law in Computer Programs" Chapter One in TJM Bench-Capon (ed.), Knowledge-Based Systems and Legal Applications (Academic Press: London, 1991) p.16. 276 S Torrance, "Breaking out of the Chinese Room" in M Yazdani (ed.), Artificial Intelligence: Principles and Applications (Chapman & Hall: London, 1986) at p.301. 277 TW Bynum, "Artificial Intelligence, Biology, and Intentional States" (1985) 16 Metaphilosophy 355. 278 T Cuda, "Against Neural Chauvanism" (1985) 48 Philosophical Studies 111. 279 DE Rumelhart, JL McClelland and the PDP Research Group, Parallel Distributed Processing: Explorations in the Microstructure of Cognition (MIT Press: Cambridge, 1986). 280 AL Tyree, "The Logic Programming Debate" (1992) 3 Journal of Law and Information Science 1111 at p.115. 281 q.v. RA Clarke, Knowledge-Based Expert Systems: Risk Factors and Potentially Profitable Application Areas(Working paper: Department of Commerce, Australian National University, 1988). 282 M Aultman, "Technology and the End of Law" (1972) 17 American Journal of Jurisprudence 46 at pp.49-52.