Japan Advanced Institute of Science and Technology

Japan Advanced Institute of Science and Technology

JAIST Repository

https://dspace.jaist.ac.jp/

Title

ムの研究

Author(s)

Citation

Issue Date

Type

Text version

URL

Rights

Description

legal analogy under incomplete knowledge

Kozo Morisawa

Scho ol of Information Science,

Japan Advanced Institute of Science and Technology

Febrary 13, 1998

Keywords: abduction, abductivelogic programming, order-sorted logic, knowledge

abstraction, clausehierarchy,legal analogy.

Inthisthesis,wedescribeamethodforinferenceofdiscoveryinlegalinference. Logical

Jurisprudenceseparateslegalinferenceintotheinferenceofjusticationandtheinference

of discovery. The inference of justication is the process to justify from the premises

regarded to true legally, and the inference of discovery is the process to discover the

judgment and premises. In Logical Jurisprudence, these inferences cannot miss in legal

inference,b ecause incase lawyersdelivera judgment,these inferencesmake upfor alack

of knowledge to execute inference in the other side. In these methods of inference, the

inference of justication can realize with deduction, however the inference of discovery

cannot realize with deduction, therefore it needs to extend a metho d for inference. In

the eld of Articial Intelligence, many studies are done concerned with the inference

of discovery using elementsto extend the inference, however,most of these studies have

several problems as regards its framework for realization of legal inference, for example,

therangeofpossibleinferencewasrestricted,oritisn'tenoughtosortoutlegalknowledge.

In order to improve these problems that past studies have, we pay attention to ab-

ductionand analogy. Abductionisa methodof inference,Peircedistinguishedab duction

fromdeduction orinduction for the rst time,he says thatab duction is the inferencefor

looking for explanations, and he also says that abduction is the process for generating

hyp otheses and explanations. Furthermore, ab duction is available to realize the infer-

ence underincomplete knowledge,Kowalskiindicatesthe p otentiality toapply ittolegal

inference. Analogy is also a method of inference, and it applys an analogous matter to

would like to know to use a well-known matter and it utilizes in legal inference, called

legal analogy. Legal analogy isa methodto interpretlegal sentence, in case there aren't

any rules to apply to the case, lawyersare trying to interpreta rule to apply the case.

Copyright c

1998byKozoMorisawa

in legal inference, we have utilized Abductive Logic Programming(ALP) system. ALP

is an extension of logic programming to realize abduction, and it is possible to realize

according to the extention of SLD resolution. We have utilized the procedure of Kakas

and Mancarella in ALP system. Furthermore, in order to realize legal analogy, we have

utilizedAbstraction-basedAnalogy. Abstraction-basedAnalogyisamethodthatassumes

anabstractdomainbetween sourcedomainand targetdomain, andweutilizedGreiner's

model inorder torealizeit. We haveintegrated thesetwoframework,that is,Abduction

and Analogy, and we have realized the inference of discovery. we have utilized Order-

sorted Logic for knowledge representationin order torealize Abstraction-based Analogy.

Order-sorted Logic is a part of sort logic, and it denes a concept of subsort. We have

utilizeditbyWalther'smodel. Basedontheseconsideration,inordertorealizeknowledge

abstraction, we utilize the sort hierarchy, and generate the clause hierarchy. In this

procedure, in order to realize knowledge abstraction, predicate name and sort attached

to clause are abstracted at the same time, as a result, this metho d realizes abstraction

by clause and forms clause hierarchy, and it is the dierence between this metho d and

past one. In the former studies of legal analogy, the method to realize abstraction was

restricted predicate name or sort attached to clause, these methods restricted the range

to realizelegal analogy.

In order to evaluate this system, we have carried out two experments, one is the

generationofknowledgehierarchy,and wehavegeneratedthe clausehierarchyusinglegal

sentences and cases by this system, the other is the legal analogy, and we have solved

legal problems, namely cases, using clause hierarchy which generated by the generation

of knowledge hierarchy. In these experments, as the law to execute the legal inference,

wehaveutilizedtheUnited NationConvention onContracts forthe InternationalSaleof

Goods(CISG),PartIIandPartII I.As thecaseknowledge,wehaveutilizedseveralcases,

thesecasesare aimedatexpermentfor legalinference,andthese casesareconcernedwith

the conclusion of the contractand the cancellationof the contract, these cases are based

onCISG.Inorder toexecutethe inference,wehavereconstructedthese knowledges,they

are translated into the expressions of Order-sorted Logic, and legal sentences translated

into rules in logic programming, and cases translated into facts. As the result of the

generation of knowledgehierarchy, inthe lack of knowledge, calledhyp othesis set inthis

thesis,thatwasgeneratedusingALPsystem,thissystemcouldutilizeahalfofhypothesis

settogenerateclausehierarchy,wehaveshowedthatthis systemcouldutilizetogenerate

clause hierarchy. As the result of the legal analogy, this system could reason a part of

hyp othesis setbyanalogy,wehaveshowedthatthissystemcouldrealizeforlegalanalogy.

In this studies, we have provided a framework to integrate ab duction and analogy,

and we couldhaveutilizedclause hierarchywhich wasgenerated to realizelegal analogy.

However,wehavethoughtthattherearesomeproblemstobeprovedinthefuture. Inthis

studies, the algorithms togenerate clause hierarchy aren'tcomplete, and there are many

partstodep endonuser'sdecisiontouse thissystem. Wehavetoexaminetheseproblems

asourfuturework. Furthermore,wehavefoundinthisstudiesthattheknowledgeswhich

needs for inference are beyond the range of legal knowledge. Based onthis discovery,we

havetoexaminethemethodforknowledgerepresentationandinferenceinordertorealize