|
Copyright 1995, KAGAYAMA Shigeru |
 |
I think that the law as a type of knowledge data base is like a computer program. If there is a bug in the system it could result in grave damages to third parties, or depending on the case, even extend to the very core of human existence. Law is a common knowledge program, the debugging of which has been a continuing process for some thousands of years already.
Up until now, this legal knowledge data base has only been accessible through the excellent minds of judges and lawyers. But lately there has been ongoing research (legal expert system research) which investigates the possibilities of automating access to parts of this legal knowledge base by utilizing the logic functions of computers. For example, it could be possible to construct a system in which the mere input of a case would call out all the relevant judgment to that particular set of facts. [1]
In this essay I would like to argue from a legal standpoint and as one who has been involved in the development of legal expert systems. My thesis is that the computerization of legal reasoning will take much the same road as the development of language translation machines has taken. By treating the process of going from pertinent legal cases to the facts, and conversely, from the facts to relevant legal cases in the same way as machine translation operates, the computerization of legal reasoning could be advanced to the point that it could be applied to more perplexing problems like getting computers to add legal meanings and significance to raw facts to reach the proper legal conclusion by utilizing the appropriate rules and case law.
[Contents] [Japanese Home Page] [Home Page]
The law is structured through an objective group of rules which are designed for the maintenance of the peace, and have as their goal the attainment of the appropriate resolution of disputes. [2]
In order for the basis of dispute resolution to be respected by the average person, more than anything it is necessary, in the case of its application that the decisions of the law be appropriate. Moreover it is necessary that the law be equal and objective so that differing results will not arise through different peoples application of it.
[Contents] [Japanese Home Page] [Home Page]
The appropriateness of dispute resolutions is actualized by pondering the flexibility of the basis of resolutions where one party to the dispute is thought of as "pitiable" and the other as "blameworthy". For example, let us consider a case of a person walking along a road in the rain and is splashed with muddy water by a passing car.
In this case, the most appropriate course of resolution would be for the driver of the car (the tortfeasor) to compensate the person who had his clothes dirtied for the cleaning expense. The reason why that would be an appropriate resolution is because, generally speaking, the victim who had his clothes dirtied is "pitiable", while the tortfeasor who should have lowered his speed to avoid spraying water should be reprimanded.
But while empathy is a convenient sentiment to help ascertain whether a resolution is appropriate or not, it is not dependable concept in terms of legal applications. There is fear that an unfair judgment would occur in a case where the tortfeasor was a starving student and the victim was a millionaire. Based on empathy alone, a decision not to compensate the millionaire because it is very difficult to feel sorry for rich people might be mistakenly arrived at. The law has to maintain objectivity and equality as well as realistic appropriateness. [3]
The realization of objectivity in dispute resolution is attained by declaring or publishing the basis for previous resolutions in the form of rules that are a mixture of relevance and validity, intertwined with the application of non technical common knowledge.
If we were to apply this kind of objectivity to the previous example where the tortfeasor was a starving student earnestly performing his part time job, and the victim was a millionaire, then it is clear that the tortfeasor would have to compensate the victim.
An example of a relevant legal rule (Article 709 of the Civil Code)
Whether intentionally or by negligence, anybody who trespasses on another's rights is responsible for the compensation for any damages which arise thereby
IF
there is a tortfeasor (whether intentional or by negligence) (A v B)Expressed in symbolic logic:
(A v B) ^ C ^ D ^ E - R[Contents] [Japanese Home Page] [Home Page]
Next let us consider a case where the driver (tortfeasor) in order to avoid a child who suddenly appears on the road, enters a puddle and splashes mud on a passer-by (victim). In this case, must the tortfeasor compensate the victim for damages in the form of cleaning expenses?
In this example, the sense of empathy towards or pitiabilty of the victim has not changed, but the tortfeasor's act has changes from something reprimandable to a praiseworthy act which was necessary to save a human life. In this case, to force the tortfeasor to carry the responsibility for the damages would be altogether too strict.
The law has imagined this kind of scenario and provided exceptions to the general rule introduced in Article 709. By joining the general principle with the following two exceptions into one modified rule and applying it to the previous example can direct us to the conclusion that it is not necessary for the tortfeasor to compensate the victim for his damages.
Exception 1:
Inability to assume responsibility - Infants and Mental Incompetents (Civil Code Article 712,713).In the case where an infant inflicts damages on another, if the child is not equipped with enough mental powers to be able to appreciate responsibility for the act, or in the case where a mentally deficient person inflicts damages on another, then in those cases the tortfeasor is not liable to carry the responsibility for the compensation of damages.
(A v B) ^ C ^ D ^ E ^ ~F - ~R (F shows ability to assume responsibility)
Exception 2:
Reasons for escape from illegality-If in order to defend one's own rights or that of a third party or, to avoid a perilous situation, the tortfeasor who had no other alternative course of action is not liable to bear the responsibility for the restitution of damages.
(A v B) ^ C ^ D ^ E ^ ~G - ~R (G shows illegality)
The amended tort principle:
Whether intentional or by negligence, if a person capable of bearing responsibility trespasses on another's rights, then he is liable to bear the responsibility for the restitution for damages which are incurred thereby.
(A v B) ^ C ^ D ^ E ^ F ^ G - R
or
(A v B) ^ C ^ D ^ E ^ ~(~F v ~G) - R
[Contents] [Japanese Home Page] [Home Page]
Legal rules show their effectiveness through their application when a dispute arises or when there is a fact pattern which holds the possibility of litigation. In the application of a legal rule to a problem, the facts of the case form the subordinate proposition while the main proposition which directs us to the resolution of the dispute is formed by the black letter law itself. The syllogistic interaction between these two proposition leads us to the resolution of the dispute. [4]
An example of a legal syllogism.
Main Proposition:
Civil Code Article 709Subordinate Proposition:
The facts (as acknowledged by the court and through standard legal interpretation)Conclusion: The judgment.
Y is responsible for the compensation of damages which he inflicted on X.
Formation of the subordinate proposition.
Main Proposition:
Legal interpretation.Subordinate Proposition:
The raw facts.Conclusion: The fact as interpreted by the court through standard
legal interpretation.
A through negligence trespassed on B's property rights and thereby damages to B arose.
[Contents] [Japanese Home Page] [Home Page]
Through the interpretation of law two objectives are accomplished. First of all, the raw fact of "not watching the road ahead while driving" would be interpreted at law as "negligence". Then by centering on the legal syllogistic mechanism, the legal rule that is most relevant to the set of facts in question can be retrieved. [5]
The retrieval of this rule is done by high level pattern matching where the facts of the disputed case are matched with the facts of the relevant judgment, thus calling out the applicable codified rules which are contained in that judgment.
[Contents] [Japanese Home Page] [Home Page]
Within the context of legal interpretation, the operation of retrieving applicable legal rules differs from simple data base retrieval in that it is not simply retrieval by associative mechanisms consisting of the specific legal name of the rule or of letter sequences of key words contained in the fact pattern. This differs from simple data retrieval in that we are working with problems grouped into similar recurring fact patterns.
Through many years of experience and inspiration legal specialists have been able to choose the most appropriate rules from the many choices applicable to a given fact pattern. However, the exact method by which they reach this conclusion is not clear. But there is no doubt that the legal result that the client would like to achieve plays a very important role in the retrieval process. In other words, instead of working from facts and legal prerequisites, legal specialists usually use the desired legal effect as a starting point in their analysis.
This differs from a mere search for facts or legal prerequisites because it consists firstly retrieving the applicable law which would yield the disputant's desired legal result and secondly testing this conclusion's appropriateness to the given situation. Legal specialists seem to perform this procedure almost unconsciously.
Compared to the amount of legal rules and prerequisites for their operation, the number of desired legal effects of a disputant are limited. Therefore by using the key words of the desired legal effects to retrieve the applicable law from the volumes existing is in fact a very effective way of retrieving the law. Of course lawyers have already mastered this system of retrieval in their minds and do not feel the need for any such computer system to assist them. This is probably one large reason why automation of legal reasoning has not been realized to a greater extent than it has up till now. [6]
[Contents] [Japanese Home Page] [Home Page]
At the heart of legal interpretation is clearly the exercise of whether the problematic fact pattern conforms to the prerequisites of the applicable legal rule. This is on type of translative process. When this translation can move from both the legal prerequisites to the facts and back again then the process will be quite automated. The first main step in legal interpretation is the judging of whether the problematic facts fall in line with ( conform to ) the prerequisites of the applicable laws.
[Contents] [Japanese Home Page] [Home Page]
The translative function that forms the bridge between the facts and the relevant case law must operate from two directions. The first operation is to make concrete sub rules from abstract legal principles.
Within the general legal principle of tort law there are concrete sections pertaining to traffic accidents, pollution, products liability, medical malpractice etc. The task of the translative operation make the rules (for each of these area) concrete through their meanings and thereby make the connection between the relevant legal case and the fact pattern. For example it might be possible to show the construction of the rules of product liability law as they have developed through the cases.
Consider the following: The construction of the requisite rules pertaining to product liability law.
Authority to seek compensation for damages. <- Product defect, Trespass to rights (infringement), Damages, Damages arise from the defective product, not (no negligence by manufacturer), not (caused by third party), not (incapable of assuming responsibility), not (escape from illegality).>
This operation is perfected to an extent by making a dictionary of possible matching. [7]
This dictionary would consist of a statistical analysis of the actual appearance of a given fact pattern within the recorded case law and examples through which the determination of which rules of which case would come under which set of facts could be made. The flow of this operation takes the form of a rule to fact top down thought process. Due to the principles of legalism, effect cannot be given to law that is not included within the body of legal rules. The contents of these rules are limited and therefore perfectly standardized. This is unlike an analysis of natural language where the rules are often prone to exceptions. This fact makes the top down mode of analysis very well suited to legal reasoning (as it is a much more predictable process when applied to law rather than the exception prone rules of natural language).
That brings us to the second operation. The fact patterns and relevant case law that have been expressed in judgments are gathered, and then by probability a single fact pattern from the relevant case law is abstracted. If it is supposed that the former operation is like an English-Japanese dictionary, then the latter would be the Japanese-English version. This operation is one that is absolutely necessary in the case of bottom up reasoning where starting from a gathering of fact patterns it is possible to determine all possible conclusions (predicted judgments). In the case of law, for reasons previously stated the top down mode of reasoning becomes the main flow, but to increase the efficiency of the reasoning, it is necessary to include the bottom up mode as well. Which ever way it is done, both these operations are like the wheels of a cart in that they must advance together in parallel.
[Contents] [Japanese Home Page] [Home Page]
When it has been determined whether or not the problematic set of facts fall under the applicable case law, and the legal interpretative process is completed, then the legal conclusion is reasoned almost automatically through the syllogistic mechanism. There however remains the question of whether this legal conclusion is appropriate or not. As expressed earlier, each litigant (disputant) has a unique set of circumstances under which light the legal conclusion should be examined. In the case of an inappropriate conclusion, the appropriateness check has to be run again after retrieving alternate law and testing alternative possible legal interpretations.
For example of proposed model of automated reasoning see APPENDIX.
[Contents] [Japanese Home Page] [Home Page]
If the standard legal rules are not able to yield an appropriate legal result then at least on the surface the judge is forced to turn to the case law to derive the appropriate rules. While this appears on the surface to be creating new rules it is in fact only abstracting law which already exists. Though this existing common law is very close to pragmatism (disposing of a case in the most appropriate way) like the rules of honor, public order and moral standards, or the general principles of tort law.
[Contents] [Japanese Home Page] [Home Page]
The special feature of legal reasoning is that it can take an unlimited amount of problematic fact patterns and yield a legal conclusion by using a limited amount of legal rules, cases and vocabulary within the reasoning process (this makes legal reasoning particularly well suited to automation). It is possible to look at the long history of judgments made by judges choosing the most appropriate resolution strategy from the limited number of legal conclusions which were derived by using a limited number of rules and legal vocabulary with respect to an unlimited amount of fact problems. Though the number and content of the rules have fluctuated slightly over time the size of the general body of rules has more or less remained consistent. Through this effort the unlimited number of fact patterns also have actually become limited through repetition and compilation into a set of publicly recorded judgments.
As for the future of automated legal reasoning, there still exist many barriers on a long and complicated road to perfection. However, with the multitude of various scholarly applications it is clear that the development of a workable automated legal reasoning system is an area of research that can no longer be ignored. The spin off benefit of the ability to predict with accuracy the result of a certain course of logic or reasoning process with respect to a given set of realistic facts will manifest itself in the form of new legal research methods. For example, a foreign country's laws could be made into a knowledge base for an automated reasoning system. Various legal problems could then be run through this system. When the same legal problems are run through an expert system based on one's own country's laws a reconstructed comparison of the two legal systems is possible. This kind of application of a legal expert system would definitely revolutionize the way comparative law is studied. Quantum leaps in the efficiency of private international law would be another beneficial by product.
In the same way, various scholarly explanations of the law could be objectively evaluated for their logical validity and realistic appropriateness. This would be done by converting the scholarly explanations into a knowledge base for the expert system. The running of various fact patterns through this system would reveal the strengths and weaknesses of the reasoning in the scholarly explanation. By applying an expert system in this way the study of law will begin to be developed into a true science.
[Contents] [Japanese Home Page] [Home Page]
I should like to thank David ANDERSSON( Law Student of University of British Columbia) who helped me for putting this thesis into English.
[Contents] [Japanese Home Page] [Home Page]
[1] H.Yoshino, et al, Horitsu Expert System no Kiso (The Basis of Legal Expert Systems) (1986).
[2] Regarding the relationship between law and rules, see J.Aomi, Ho to Shakai: Atarashii Hogaku Nyumon (Law and Society: A New Entrance To Legal Studies) (1967). Also note Y.Matsuura, Common Law no Hi-ruruteki na Shiko no Saisei (The Common Law and the Resurgence of Casuistry) Nos.145,146 (1988) Osaka L.Rev.299.
[3] Within circles of Legal Scholars two rivaling schools of thought can be found as to which should receive more emphasis- realistic appropriateness or objectivity. For a more detailed explanation see Hogaku Bunken Senshu 1. Hogaku no Hoho (Anthology of Legal Literature 1.The Methods of Law) (1972).
[4] For details about legal syllogisms see H.Yoshino, Hoteki Suiron no Kihon Kozo- Jittai Ho Suiron to sono Shisutemuka no Hoho (The Basis Structure of Legal Reasoning- The Systematization of Actual Legal Reasoning) included in H.Yoshino, et al, Horitsu Expert System no Kiso (The Basis of Legal Expert Systems) (1986) 105.
[5] Easier said than done. Presently, March 3rd, 1988 in Japan there are 1,556 legal codes which when combined with ordinance and regulation code the number swells to 5,872 codes. All these legal rules are recorded in the 11 volumes of the Genko Nihon Hoki (Presently enacted legal rules of Japan). The total number of pages in these 11 volumes is 135,000. For details see H.Ishikawa, Horeihensan Jigyo no Genjyo to Tenbo (The Present and Future Enterprise OF COMPILING The Laws and Ordinances). The author also confirmed by phone with the Japanese Ministry of Justice that the above figures are the most recent.
[6] Regarding this point, if the automatic retrieval of laws and ordinances for legal interpretation is to be realized, a system for retrieving laws by legal effect will need to be developed rather that just retrieving laws by name or legal prerequisite.
[7] With respect to the dictionary see the author's article, Horitsu Expert System no Jiten (Dictionary for Legal Expert Systems) included in H.Yoshino, et al, Horitsu Expert System no Kiso (The Basis of Legal Expert Systems) (1986) 142. M.Nagao, Kikai Honyaku wa doko made Kano ka (How Far is Machine Translation Possible) (1986) 88. H.Uchida, Denshika Jiten no Hatten (The Development of Electronic Dictionaries) Joho Shori Gakkai- Shizen Gengo Shori Gijutsu Shimpojiumu Hokokusho (Natural Linguistic's Technical Disposition Symposium Report) (1988) 89.
[Contents] [Japanese Home Page] [Home Page]
A model of automated reasoning in tort case
User : ?-run.
Machine : Did the tortfeasor knowing that damages would arise act purposely?
User : why
Machine : Reason for the question : It is necessary for victim to prove tortfeasor's
intention or negligence in order to claim compensation for damages.
Machine : Did the tortfeasor knowing that damages would arise act purposely?
User : why
Machine : Reason for the question : Article 709 of Civil Code of Japan requires for
a victim to prove intention or negligence of a tortfeasor.
Machine : Did the tortfeasor knowing that damages would arise act purposely?
User : why
Machine : Reason for the question has been fully explained!
Machine : Did the tortfeasor knowing that damages would arise act purposely?
User : no
Machine : In spite of being in a position to foresee damages would arise,
did the tortfeasor fail to make efforts to evade damages?
User : yes
Machine : Was there a trespass to the victim's right?
User : yes
Machine : Did damages to the victim arise?
User : yes
Machine : Was there a causal connection between the tortfeasor's action and the victim's
damages?
User : yes
Machine : Is the tortfeasor a child or mental incompetent unable to appreciate the nature
or quality of his act?
User :why
Machine : Reason for the question : Legal capacity of tortfeasor to assume responsibility
is necessary for the victim to claim compensation for damages.
Machine : Is the tortfeasor a child or mental incompetent unable to appreciate the nature
or quality of his act?
User : why
Machine : Reason for the question : According to Article 712, 713 of Civil Code of Japan
a victim can not claim compensation for damages to tortfeasor who is unable to
assume responsibility.
Machine : Is the tortfeasor a child or mental incompetent unable to appreciate the nature
or quality of his act?
User : why
Machine : Reason for the question has been fully explained!
Machine : Is the tortfeasor a child or mental incompetent unable to appreciate the nature
or quality of his act?
User : I don't know.
Machine : Was the tortfeasor's act a necessary act of self defence?
User : why
Machine : Reason for the question : Illegality in the tortfeasor's action is necessary
for victim to claim compensation for damages.
Machine : Was the tortfeasor's act a necessary act of self defence?
User : why
Machine : Reason for the question : According to Article 720 of Civil Code of Japan
a victim can not claim compensation for damages to a tortfeasor who acted for
self-defense or for avoiding danger
in case of emergency.
Machine : Was the tortfeasor's act a necessary act of self defence?
User : I don't know.
Machine : Result of legal reasoning : It is proved for the victim to claim for damages
to the tortfeasor.
User : how
Machine : The proof of the top goal(Authority for the compensation of damages based on tort)
is completed.
Requirement of Art. 709 is sufficiently proved and Requirement of Art. 712, 713, 720
obstacle of the authority is not proved sufficiently.
Machine : Result of fact finding :
intention: negative
negligence: affirmative
trespass to right: affirmative
damages arose: affirmative
causal relationship: affirmative
inability: unknown
no illegality: unknown
[Contents] [Japanese Home Page] [Home Page]
|
Copyright 1995, KAGAYAMA Shigeru |
|
