Artificial Intelligence and Legal Analytics: New Tools for Law Practice in the Digital Age

451 Pages·2017·39.16 MB·English

by  Kevin D. Ashley

“FM” — 2017/5/31 — 10:36 — page i — #1
artificial intelligence and legal analytics
NewToolsforLawPracticeintheDigitalAge
Thefieldofartificialintelligence(AI)andthelawisonthecuspofarevolutionthatbegan
withtextanalyticprogramslikeIBM’sWatsonandDebaterandtheopen-sourceinforma-
tionmanagementarchitecturesonwhichtheyarebased.Today,newlegalapplications
arebeginningtoappear,andthisbook–designedtoexplaincomputationalprocesses
to non-programmers – describes how they will change the practice of law, specifically
byconnectingcomputationalmodelsoflegalreasoningdirectlywithlegaltext,generat-
ingargumentsforandagainstparticularoutcomes,predictingoutcomes,andexplaining
thesepredictionswithreasonsthatlegalprofessionalswillbeabletoevaluateforthem-
selves.Theselegalappswillsupportconceptuallegalinformationretrievalandenable
cognitivecomputing,enablingacollaborationbetweenhumansandcomputersinwhich
eachperformsthekindsofintelligentactivitiesthattheycandobest.Anyoneinterested
inhowAIischangingthepracticeoflawshouldreadthisilluminatingwork.
Dr. Kevin D. Ashley isaProfessorofLawandIntelligentSystemsattheUniversityof
Pittsburgh,SeniorScientist,LearningResearchandDevelopmentCenter,andAdjunct
Professor of Computer Science. He received a B.A. from Princeton University, a JD
fromHarvardLawSchool,andPh.D.incomputersciencefromtheUniversityofMas-
sachusetts. A visiting scientist at the IBM Thomas J. Watson Research Center, NSF
Presidential Young Investigator, and Fellow of the American Association for Artificial
Intelligence, he is co-Editor-in-Chief of Artificial Intelligence and Law and teaches in
the University of Bologna Erasmus Mundus doctoral program in Law, Science, and
Technology.
Downloaded from https://www.cambridge.org/core. Faculty of Classics, University of Cambridge, on 03 Nov 2017 at 08:54:17, subject to the
Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316761380
“FM” — 2017/5/31 — 10:36 — page ii — #2
Downloaded from https://www.cambridge.org/core. Faculty of Classics, University of Cambridge, on 03 Nov 2017 at 08:54:17, subject to the
Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316761380
“FM” — 2017/5/31 — 10:36 — page iii — #3
Artificial Intelligence and Legal Analytics
new tools for law practice in
the digital age
KEVIN D. ASHLEY
UniversityofPittsburghSchoolofLaw
Downloaded from https://www.cambridge.org/core. Faculty of Classics, University of Cambridge, on 03 Nov 2017 at 08:54:17, subject to the
Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316761380
“FM” — 2017/5/31 — 10:36 — page iv — #4
UniversityPrintingHouse,CambridgeCB28BS,UnitedKingdom
OneLibertyPlaza,20thFloor,NewYork,NY10006,USA
477WilliamstownRoad,PortMelbourne,VIC3207,Australia
4843/24,2ndFloor,AnsariRoad,Daryaganj,Delhi–110002,India
79AnsonRoad,#06-04/06,Singapore079906
CambridgeUniversityPressispartoftheUniversityofCambridge.
ItfurtherstheUniversity’smissionbydisseminatingknowledgeinthepursuitof
education,learning,andresearchatthehighestinternationallevelsofexcellence.
www.cambridge.org
Informationonthistitle:www.cambridge.org/9781107171503
doi:10.1017/9781316761380
©KevinD.Ashley2017
Thispublicationisincopyright.Subjecttostatutoryexception
andtotheprovisionsofrelevantcollectivelicensingagreements,
noreproductionofanypartmaytakeplacewithoutthewritten
permissionofCambridgeUniversityPress.
Firstpublished2017
PrintedintheUnitedStatesofAmericabySheridanBooks,Inc.
AcataloguerecordforthispublicationisavailablefromtheBritishLibrary.
LibraryofCongressCataloging-in-PublicationData
ISBN978-1-107-17150-3Hardback
ISBN978-1-316-62281-0Paperback
CambridgeUniversityPresshasnoresponsibilityforthepersistenceoraccuracy
ofURLsforexternalorthird-partyinternetwebsitesreferredtointhispublication
anddoesnotguaranteethatanycontentonsuchwebsitesis,orwillremain,
accurateorappropriate.
Downloaded from https://www.cambridge.org/core. Faculty of Classics, University of Cambridge, on 03 Nov 2017 at 08:54:17, subject to the
Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316761380
“FM” — 2017/5/31 — 10:36 — page v — #5
ForAlida,forever
Downloaded from https://www.cambridge.org/core. Faculty of Classics, University of Cambridge, on 03 Nov 2017 at 08:55:23, subject to the
Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316761380
“FM” — 2017/5/31 — 10:36 — page vi — #6
Downloaded from https://www.cambridge.org/core. Faculty of Classics, University of Cambridge, on 03 Nov 2017 at 08:55:23, subject to the
Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316761380
“FM” — 2017/5/31 — 10:36 — page vii — #7
Contents
Listofillustrations pagexv
Listoftables xxi
Acknowledgments xxiii
parti computationalmodelsoflegalreasoning 1
1 IntroducingAI&LawandItsRoleinFutureLegalPractice 3
1.1. Introduction 3
1.2. AI&LawandthePromiseofTextAnalytics 4
1.3. NewParadigmsforIntelligentTechnologyinLegalPractice 6
1.3.1. FormerParadigm:LegalExpertSystems 8
1.3.2. AlternativeParadigms:ArgumentRetrievalandCognitive
Computing 11
1.3.3. TowardtheNewLegalApps 14
1.4. WhatWatsonCanandCannotDo 14
1.4.1. IBM’sWatson 15
1.4.2. QuestionAnsweringvs.Reasoning 18
1.4.3. IBM’sDebaterProgram 23
1.4.4. TextAnalyticToolsforLegalQuestionAnswering 27
1.4.5. SourcesforTextAnalyticTools 30
1.5. AGuidetoThisBook 31
1.5.1. PartI:ComputationalModelsofLegalReasoning 32
1.5.2. PartII:LegalTextAnalytics 33
1.5.3. PartIII:ConnectingComputationalReasoningModels
andLegalTexts 34
1.6. ImplicationsofTextAnalyticsforStudents 35
vii
Downloaded from https://www.cambridge.org/core. University of Florida, on 03 Nov 2017 at 09:05:57, subject to the Cambridge Core terms
of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316761380
“FM” — 2017/5/31 — 10:36 — page viii — #8
viii Contents
2 ModelingStatutoryReasoning 38
2.1. Introduction 38
2.2. ComplexitiesofModelingStatutoryReasoning 39
2.2.1. SemanticAmbiguityandVagueness 39
2.2.2. SyntacticAmbiguity 40
2.3. ApplyingStatutoryLegalRulesDeductively 42
2.3.1. RunningaNormalizedVersiononaComputer 43
2.3.2. PredicateLogic 44
2.3.3. SyntacticAmbiguityasDesignConstraint 45
2.3.4. TheBNAProgram 47
2.3.5. SomeProblemsofTranslatingStatutesintoPrograms 48
2.4. TheComplexityofStatutoryInterpretationandtheNeedfor
Arguments 52
2.4.1. AStepwiseProcessofStatutoryInterpretation 53
2.4.2. OtherSourcesofLegalIndeterminacy 54
2.5. ManagementSystemsforBusinessRulesandProcesses 56
2.5.1. BusinessProcessExpertSystems 56
2.5.2. AutomatingBusinessProcessCompliance 60
2.5.3. RequirementsforaProcessComplianceLanguage 62
2.5.4. ConnectingLegalRulesandBusinessProcesses 65
2.5.5. ExampleofBusinessProcessComplianceModeling 68
2.6. RepresentingStatutoryNetworks 70
3 ModelingCase-basedLegalReasoning 73
3.1. Introduction 73
3.2. RelationshipofLegalConceptsandCases 74
3.2.1. TheLegalProcess 74
3.2.2. TheLegalProcessIllustrated 75
3.2.3. RoleofLegalConcepts 76
3.3. ThreeComputationalModelsofLegalConceptsandCases 77
3.3.1. PrototypesandDeformations 78
3.3.2. DimensionsandLegalFactors 81
3.3.3. Exemplar-basedExplanations 93
3.4. TeleologicalModelsofCase-basedLegalReasoning 97
3.5. AnApproachtoModelingTeleologicalReasoning 100
3.5.1. TeleologyinTheoryConstruction 101
3.6. DesignConstraintsforCognitiveComputingwithCase-based
ModelsofLegalReasoning 104
4 ModelsforPredictingLegalOutcomes 107
4.1. Introduction 107
4.2. ANearestNeighborApproachtoAutomatedLegalPrediction 108
Downloaded from https://www.cambridge.org/core. University of Florida, on 03 Nov 2017 at 09:05:57, subject to the Cambridge Core terms
of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316761380
“FM” — 2017/5/31 — 10:36 — page ix — #9
Contents ix
4.3. IntroductiontoSupervisedMachineLearning 109
4.3.1. MachineLearningAlgorithms:DecisionTrees 110
4.4. PredictingSupremeCourtOutcomes 111
4.4.1. FeaturesforPredictingSupremeCourtOutcomes 112
4.4.2. Applying Supervised Machine Learning to SCOTUS
Data 112
4.4.3. EvaluatingtheMachineLearningMethod 113
4.4.4. MachineLearningEvaluationMeasuresandResults 114
4.5. PredictingOutcomeswithCase-basedArguments 114
4.5.1. PredictionwithCATO 115
4.5.2. Issue-basedPrediction 115
4.5.3. IBP’sPredictionAlgorithm 117
4.5.4. EvaluatingIBP’sPredictions 119
4.6. PredictionwithUnderlyingValues 121
4.7. PredictionbasedonLitigationParticipantsandBehavior 123
4.8. PredictioninCognitiveComputing 125
5 ComputationalModelsofLegalArgument 127
5.1. Introduction 127
5.1.1. AdvantagesofCMLAs 128
5.2. TheCarneadesArgumentModel 129
5.3. AnExtendedExampleofaCMLAinAction 131
5.3.1. FamilyLawExamplewithCarneades 132
5.3.2. ArguingwithDefeasibleLegalRules 134
5.3.3. IntegratingArguingwithCasesandRules 135
5.4. ComputationalModelofAbstractArgumentation 139
5.5. HowCMLAsComputeWinnersandLosers 141
5.5.1. ResolvingConflictingArgumentsaboutFacts 142
5.5.2. ResolvingConflictingArgumentsaboutValues 143
5.5.3. ResolvingConflictingArgumentsaboutLegalRules 144
5.6. How Practical are Computational Models of Legal
Argument? 144
5.6.1. RoleofProofStandardsinCMLAs 145
5.6.2. IntegratingProbabilisticReasoningintoCMLAs 147
5.7. ValueJudgment-basedArgumentativePredictionModel 149
5.7.1. VJAPDomainModel 150
5.7.2. VJAPValuesUnderlyingTradeSecretRegulation 151
5.7.3. VJAPArgumentSchemes 154
5.7.4. VJAP’sArgument-basedPredictions 156
5.7.5. VJAPProgramEvaluation 158
5.8. ComputationalModelofEvidentiaryLegalArgument 160
5.9. ComputationalModelsofLegalArgumentasaBridge 164
Downloaded from https://www.cambridge.org/core. University of Florida, on 03 Nov 2017 at 09:05:57, subject to the Cambridge Core terms
of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316761380