Semantics In The Legal Domain

1
Semantics In The Legal Domain

• By
• Angela Maduko

2
Introduction

• This presentation looks into
• Different ontologies in the legal domain and the
  motivations for using using ontologies in legal
  information/knowledge systems
• Applications of some of the legal ontologies and
  a comparison of two such ontologies
• Metadata standards in the legal domain.
• Some tools and applications of the legal
  ontologies/metadata.

3
Overview of KBS in the Legal Domain

• Original Concept was
• A set of empirical associations formed by an
  expert in the course of practising in the domain
• This is encoded in some executable formalism
  typically production rules
• Problems
• Knowledge Acquisition Experts often had
  difficulties in making their expertise explicit
  and stating their empirical associations with
  sufficient completeness
• Robustness The systems were not able to deal
  with situations not anticipated/overlooked by the
  experts
• Maintenance and reuse Was built for a particular
  task and so was difficult to adapt to other tasks
• Explanation Poor quality explanations since the
  empirical associations did not reflect the
  underlying causal mechanisms of the domain

4
Need for Ontology

• Knowledge Sharing
• Verification of a Knowledge base
• Software Engineering considerations
• Knowledge Acquisition
• Knowledge Re-use
• Domain theory development

5
Need for Ontology

• A simple example
• UK SS laws concept of pensionable age 65
  (man),
• 60 (woman)
• c1, pensionable_age(X),- sex(X, male), age(X,
  A), A,gt 65.
• c2, pensionable_age(X),- age(X, A), A,gt 60.
• This can also be represented as
• c1, pensionable_age(X),- sex(X, male), age(X,
  A), A,gt,64.
• c2, pensionable_age(X),- sex(X, female), age(X,
  A), A,gt,59.

6
Building IR oriented Legal Ontologies

• Issues
• Identifying domain terms
• Lexical phenomenon such as synonymy, polysemy
• Controversies existing among experts
• Some methods are
• Top-down method
• Specifying or generalizing an existing ontology
  to create a new one
• Agreement on a unique point of view on the
  specialties of domain experts, which becomes the
  basis of the new ontology
• Bottom-up method
• Extracting all elements needed to build an
  ontology from appropriate documents
• Selecting an appropriate corpus
• Extracting domain terms
• Identifying relations among terms

7
Some Legal Ontologies

• Valentes Functional Ontology of Law
• Frame-based Ontology of Van Kralingen and Visser
• McCartys Language of Legal Discourse(LLD)
• Stampers Norma

8
Legal Ontologies

• Functional Ontology of law - based on a
  functional perspective of the legal system.
• Main function - reacting to social behaviour - is
  decomposed into six primitive functions
• Normative Knowledge
• World Knowledge
• Responsibility Knowledge
• Reactive Knowledge
• Meta-legal Knowledge
• Creative Knowledge

9
Legal Ontologies

• Frame-based Ontology - decomposed into
• Generic legal ontology
• Norms
• Acts
• Concept descriptions
• Statute-specific ontology

10
Frame-based Ontology

• Norms
• Norm identifier
• Norm type
• Promulgation
• Scope
• Conditions of application
• Norm subject
• Legal modality
• Act identifier

11
Frame-based ontology

• Act identifier
• Promulgation
• Scope
• Agent
• Act type
• Modality of means
• Modality of manner

• Temporal aspects
• Spatial aspects
• Circumstantial aspects
• Cause of action
• Aim of action
• Intentionality of action
• Final state

Acts (Events, Processes institutional acts,
Physical acts)
12
Frame-based ontology

• Concept descriptions comprise
• Concept to be described
• Concept type (definition, deeming provision,
  factor, meta)
• Priority
• Promulgation
• Scope
• Conditions of application
• Enumeration of instances

13
Comparison of the two ontologies

• Criteria for comparison
• Epistemological adequacy (clarity, intuitiveness,
  relevance, completeness, discriminative power)
• Operationality (Encoding bias, coherence,
  computationality)
• Reusability (task-and-method reusability, domain
  reusability)

14
Some Applications of legal Ontologies

• Functional ontology of law
• ON-LINE (Ontology-based Legal Information
  Environment)
• Frame-based ontology
• An assessment expertise system( Dutch
  Unemployment Benefits Act(DUBA) )
• A planning expertise system
• CLIME ontology(Cooperative Legal Information
  Management and Explanation)
• Maritime Information and Legal Explanation(MILE)
• Knowledge Desktop Environment(KDE)

15
ON-LINE

• An architecture for storing and retrieving legal
  information
• and reasoning with legal knowledge
• Based on the idea that Legal problem solving is
  to some extent a
• global modelling activity in which the
  practitioner experiments with
• alternative models(interpretations) of the
  legislation and/or of a
• case in order to reason about their consequences
• Contains reasoning modules based on some models
  for legal
• assessment and legal planning as well as on
  supporting tools for
• legal modelling and design

16
ON-LINE

• Major functions
• Legal Information Serving
• Legal Analysis
• Main features
• Integrated storage and representation of legal
  text and knowledge by using interconnected
  knowledge and text repositories
• Representation of legal knowledge based on the
  functional ontology of law
• Emphasis on legal modelling as a central task in
  legal practice

17
ON-LINE (Structure)
Source ON-LINE An Architecture for Modelling
Legal Information
18
ON-LINE(Modelling Links)

• Enables the explicit storage of the modelling
  process
• Definitional links These keep track of
  information source that has been used for
  modelling a certain element of the knowledge base
• Referential links These keep track of multiple
  references to the same defined concepts

19
ON-LINE
Source ON-LINE An Architecture for Modelling
Legal Information
20
ON-LINE(Legal Information Server)

• Services directly related to Legal Information
  textual base
• Similar design with legal information databases
• Can also search for information using the
  knowledge base
• For eg, one can search for the word software in
  both the textual base and the knowledge base
  thereby enabling conceptual retrieval
• Search is made using elements in the whole
  ontology
• Retrieval is not only by queries but also based
  on the description(in knowledge terms) of a case.

21
ON-LINE(Legal Information Modelling Toolkit)

• Contains a number of tools(different browsers and
  editors) for modelling legal information
• The browsers present different view of the
  elements in the text and knowledge bases
• The editors enable adding to and deleting from
  the knowledge base
• Contains the tools for creating and managing the
  modelling links.

22
ON-LINE(Legal Analysis Environment)

• Contains an extensible number of modules that
  execute legal reasoning tasks. Two currently
  supported ones are
• Legal Assessment task accesses a case
  description(a description of relevant facts in
  the world) based on a body of legal knowledge.
• Case analysis mode A specific case already
  modelled and stored in the system is matched
  against a knowledge base
• Goal oriented mode here conditions which are
  sufficient to warrant a certain(desired)
  conclusion is sought.
• Legal Planning task generates a plan aimed at
  achieving a certain legal goal(defined in terms
  of legal concepts and norms which apply in the
  final state) starting from an initial state

23
ON-LINE

• Some problems and limitations
• Architecture is modelling intensive
• Most of this modelling work has to be
  done/checked by a specialist
• The scope of the architecture is restricted to
  model in detail limited amounts of legislation
• Epistemological intuitiveness of the ontology

24
A Legal Ontology tool (LODE)

• Gets an initial ontology from a user and refines
  it using two ontologies (General and Case
  ontologies)
• Two main issues involved
• Determining the best correspondence to a given
  concept in the general/case ontology
• Corrects bugs(missing concepts, existing
  unnecessary concepts, flawed hierarchical
  relationships etc) in the initial legal ontology
  using the extracted concepts
• EDR Electronic Dictionary serves as the general
  ontology
• With an extended version of the Sort Taxonomy
  Tool, the case ontology is built

25
LODE (EDR Electronic Dictionary)
Source LODE A Legal Ontology Development
Environment
26
LODE (Sort Taxonomy Tool)

• Builds a sort taxonomy, based on facts input by
  the user
• Sort - set of terms that occur in the same
  argument places of the same predicates
• Class a sort or set of sorts that have the same
  set of terms
• The most general class is all, and there could be
  subclasses and intersection classes
• For example With an input of the facts
• ownership(a, b).
• ownership(c, d).
• STT would generate class_1a,c and class_2b,d
• Then when fact country(b) is input, it creates
• class_3b as a subset of class_2b,d

27
LODE (SST)
Source LODE A Legal Ontology Development
Environment
28
LODE (Extended SST)

• Creating some sub-nodes of USS when the same USS
  is assigned to different arguments of predicates
• Sort taxonomy process is applied to arguments of
  functions that are arguments of predicates

Source LODE A Legal Ontology Development
Environment
29
LODE(Algorithms used for matching)

• For general ontology
• input name and definitions of a legal concept
• output the small space that can have the best
  correspondence in the concept dictionary
• Spell match - against the word dictionary
• Finds the lower boundary of the small space
• Definition match - against the concept dictionary
• Finds the upper boundary of the small space
• The space between the lower and upper boundaries
  of the concept dictionary is then extracted
• User selects best correspondence to the given
  legal concept
• For case ontology
• User finds out the best correspondence of a given
  legal concept in a case ontology

30
LODE

• Static Analysis
• Comparison of the number of immediate sub-nodes
• Distance from a root to a concept
• Topological relations between two concepts in
  each hierarchy
• Concept definitions

31
LODE
Source LODE A Legal Ontology Development
Environment
32
LODE
Source LODE A Legal Ontology Development
Environment
33
Metadata in the Legal Domain - Benefits

• To owners of legal websites
• Accurate metadata means that all information on a
  particular topic is readily accessible
• It provides a clear and consistent structure for
  the storage of information
• It promotes regular maintenance of the data
  through the identification of data that has not
  been updated since a certain date
• It offers indirect evidence of the quality of the
  data in that an organisation that invests time
  and money in the creation of accurate metadata is
  likely to have made a similar investment in the
  data itself
• It increases the visibility of the website to
  Search Engines
• It increases the acceptability of the website to
  Search Engines

34
Benefits of metadata

• To users of legal websites
• Searching for information is easier and more
  effective with consistent terminology
• It allows for an increase in precision and recall
• It allows for an indirect assessment to be made
  of the quality of the information.

35
Justice Sector Metadata Standard

• Based on Australian Government Locator Service
  (AGLS)
• Designed for organizations publishing legal
  materials on the web in NSW
• Objectives include
• To improve quality of access
• To reduce costs

36
Justice Sector Metadata Standard
Standard metadata fields for the Justice Sector
are

• DC.Title
• DC.Creator
• DC.Publisher
• DC.Rights
• DC.Subject
• Keywords
• DC.Description
• Description
• DC.Language
• DC.Coverage
• DC.Coverage.Jurisdiction

• DC.Coverage.Region
• DC.Date.Created
• DC.Date.Modified
• JSMS.Category
• DC.Type
• DC.Format
• DC.Identifier
• AGLS.availability
• Admin.Creator
• Admin.DateCreated
• Admin.DateValidTo

37
Legal and Advice Sectors Metadata Standard
Based on Dublin Core The elements include

• Mandatory Elements
  
• Title
  
• Author
  
• Subject
• Description
• Publisher
• Date.created
• Date.modified
• Resource identifier
• Language
• Coverage

• Optional Elements
• Contributor
• Resource type
• Format
• Source
• Relation
• Rights management

38
Some applications of Legal Metadata

• EULEGIS (European User Views to Legislative
  information in Structured Form)
• Legal RDF Dictionary
• Lawzone (a metadata enabled search facility)

39
EULEGIS

• Purpose Providing an integrated access to the
  numerous European legal databases
• Goal Improving information retrieval through the
  use of structured documents
• Implementation Relational XML-based metadata
  database containing data about legal systems and
  legal databases

40
EULEGIS

• Modules of the EULEGIS metadata model

Database
Documents
Actors
Process
Source XML Metadata for Accessing Heterogeneous
Legal Databases
41
EULEGIS

• Legal database metadata
• Main functions
• Formation of a unified interface for querying all
  databases
• Unifying the result of the query so as to appear
  similar to the user, the original database
  notwithstanding
• Includes
• Query interfaces
• Query fields
• Allowed operators
• Informative content specified in different
  languages to make multi-lingual

42
EULEGIS

DTD fragment for describing a query interface
DTD fragment for describing query fields
Source XML Metadata for Accessing Heterogeneous
Legal Databases
43
EULEGIS

• Legal Actors metadata

Source XML Metadata for Accessing Heterogeneous
Legal Databases
44
EULEGIS

• Document types metadata

DTD fragment for describing document types
Source XML Metadata for Accessing Heterogeneous
Legal Databases
45
EULEGIS

• Legal processes metadata

DTD fragment for describing legal processes
Source XML Metadata for Accessing Heterogeneous
Legal Databases
46
EULEGIS

• Metadata visualisations
• Actor view
• Information sources view
• Process view
• Acessing Legal data
• Graphical views
• Choosing one or more databases

47
Legal RDF Dictionary

• Concept of legal RDF dictionary Maps one
  datastructure DTD or
• XML schema to another to make them comparable and
• exchangeable, thereby declaring semantics of DTDs
  or XML
• schema
• Goal To facilitate cultural diversity in the
  standardization process of
• the legal domain thereby taking advantage of the
  possibility of XML
• to create one global legal information space,
  allowing for diversity
• at the same time
• Maps key legal terms across language and
  jurisdiction borders,
• obviating the problem of literally translating
  legal terms from one
• language to the other
• The concept is applicable at many levels

48
Legal RDF Dictionary

• Mapping a term from a document to another
  jurisdiction(German) occurs
• as follows
• Establish the DTD/Schema of the document
• Establish the interface of the DTD/Schema at the
  RDF Dictionary
• Establish which interfaces have linked German
  DTDs/Schemas to the same archetypes
• Establish the German term which has the same
  archetypes
• Legal RDF Dictionary Projects
• LEXML - multi-lingual and multi-jurisdictional
  rdf dictionary for the legal world (latest
  version)
• LegalXML - English language legal terms
• European Legal RDF Dictionary

49
Some Legal Metadata Tools

• Justice Sector Metadata Html Generator
• Legal and Advice Sectors Metadata editor

50
References

• Robert Van Kralingen A Conceptual Frame-Based
  Ontology for the Law
• Joost Breuker, Andre Valente and Rabboud Winkels
  Legal Ontologies A Functional View
• Trevor J.M. Bench-Capon and Pepijn R.S. Visser
  Ontologies in Legal Information Systems The Need
  for Explicit Specifications of Domain
  Conceptualisations
• Murk Muller Legal RDF Dictionary
• Andre Valente and Joost Breuker ON-LINE An
  Architecture for Modelling Legal Information
• Chizuru Aoki, Masaki Kurematsu and Takahira
  Yamaguchi LODE A Legal Ontology Development
  Environment
• Pepijn R. S. Visser and Trevor J.M. Bench-Capon
  A Comparison of Two Legal Ontologies
• Guiraude LAME Constructing an IR-Oriented Legal
  Ontology
• T.J.M. Bench-Capon and P.R.S. Visser Deep
  Models, Ontologies and Legal Knowledge Based
  Systems
• Virpi Lyytikainen, Pasi T. Tiitinen and Airi
  Salminen XML Metadata for Accessing
  Heterogenous Legal Databases
• http//www.lcd.gov.uk/consult/meta/metafr.htmpart
  6 Metadata Scheme for Websites in the Legal and
  Advice Sectors
• http//www.agd.nsw.gov.au/agd.nsf/pages/lawzone
  LawZone A new Way of Searching