Erich Schweighofer University of Vienna, Austria.

Erich Schweighofer  University of Vienna, Austria

Outline (1)State of the art: legal research and legal

retrieval systems (TXT, HTML) + hypertext, + some meta informationN-Lex: Standard for exchange of legal

information Good start, but improvements necessary

Legal Semantic Web, Legal Social WebXML, RDF, RDF schema, OWL Knowledge management in legal units

Known applications: knowledge representation, conceptual information retrieval, advanced lexical thesauri, exchange standards (MetaLex)

2

Outline (2)Other uses; more support for European

legal work?Status quo of legal searching insufficientExchange of electronic legal meta data a big

problem Need for a legal „Dublin Meta Core“

Future: Dynamic Electronic CommentarySupport tools for European legal work

Next stepsConclusions

3

Text archive & retrieval (1)

Standard service Easy access and efficient handling of the now

so many legal documents Retrieval: discrimination task more and more

difficult (e.g. finding the Boolean combination that sufficiently selects only those documents I am interested in [e.g. finding 1 to 10 documents in a collection of 1 to x million documents])

Text archive & retrieval (2)Legal retrieval ≠ “To Google”

Exact legal provision (or paragraph in a legal judgement); not just some information available in a redundant way

No Social Web (e.g. lawyers as a community are linking sufficiently to important legal documents)

Only in law firms with efficient knowledge management possible

Semantic Web?

5

Semantic WebTim Berners-Lee:

[T]he Semantic Web is "not a separate Web but an extension of the current one, in which information is given well-defined meaning, better enabling computers and people to work in cooperation”.

Standards for semantic information on the web Tagged and linked using the technologies of Resource

Description Framework (RDF), XML and URIs Web Ontology Language (OWL) Next layer: may be a logical one, an inference

machineRemains largely unrealised

6

Legal web, legal text corpora and beyondLegal web = huge text corpora

Legal information systemsWeb sites

Boolean logic + hypertextSome mark-up (text structure)

Good coverage, easy handling of documentsProblem: semantic meaning and searching is

insufficiently developed Same situation as semantic web To do: adaptation of standards for mark-up +

implementation

7

8

XML (eXtensible Markup Language)XML (eXtensible Markup Language)

General specification for creating markup languages

Subset of the Standard Generalized Markup Language (SGML) but human-legible

Free standard of the W3C; versions 1.0 and 1.1 Recommendation XML 1.1 (Second Edition):

http://www.w3.org/TR/2006/REC-xml11-20060816/ Structure has to be represented like a tree Document type definition (DTD)

Mark-up tags are freely extensible Allows semantic mark-up Law: definition of semantic document structure,

e.g.: <!ELEMENT judgement (title, summary, grounds, operational part, citations*)>

Attribute values Automatic verification

XML (2)Valid documents conform with a particular

DTD/schemaXML schema definition (XSD)

Successor of DTDsXML Style Sheet

Extensible Style Sheet Language (XSL) Client-side XSLT XML-based document transformation language

Extensible Linking and Pointer Languages XLink: simple and multiple linksXpointer: links to other document parts

Browser: Internet Explorer from version 5.0File format: OpenOffice, Word2007DTD for legal documents for Electronic Data

Interchange (EDI)

9

10

Why XML?Why XML?Advantages

Semantic meaning for syntactic data <name>schweighofer</name> Reuse + recycling of information Change of layout Improved searching of documents Unicode Open document format

Disadvantages Hierarchical model for representation has its limits Redundancy of data

Main uses in law Interchange of documents Interchange of knowledge

11

12

13

RDF (Resource Description RDF (Resource Description Framework)Framework)RDF syntax

Description of meta data in web documentsEach data can be linked with a file that describes the type of

this data. Recommendation: http://www.w3.org/TR/2004/REC-rdf-

primer-20040210/ RDF statement: subject-predicate-object expression

(triples in RDF terminology) Subject (described websource = URL) – predicate (attribute,

e.g. author) – object (value, e.g. name)Query language for RDF graphs: SPARQL Semantic web

Automated storage, exchange and use of machine-readable information on the web

Applications: exchange and common use of web data, improved implementation of search engines, classification of a website (also with software agents) etc.

14

RDF SchemaRDF SchemaExtensible knowledge representation languageLanguage for description of the structure, the content

and the semantic of XML documentsBasic elements for the description of ontologies (RDF

vocabularies) Recommendation: 10.2.2004:

http://www.w3.org/TR/2004/REC-rdf-schema-20040210/A RDF scheme does not only describe predicates of a web

source (e.g. title, author, etc.), but also the kind of the described sources (e.g. books).

Development of user-oriented RDF vocabulariesObject-oriented description of data structures with

multiple heritageClasses, predicates, constraints

Ontology for exchange of data on the WebImportant initiatives: Dublin Core Metadata Initiative,

PICS labels, P2P

Web Ontology Language (OWL)Family of knowledge representation languages

for developing ontologiesRevision of the DAML+OIL web ontology language W3C standard 10 February 2004

http://www.w3.org/2004/OWL/Two semantics based on Description Logics

OWL-DL All OWL language constructs

OWL-Lite Classification hierarchy and simple constraints (not widely used)

RDF/XML syntaxLKIF Core Ontology (Leibniz Center for Law,

Amsterdam)

15

Semantic Web & ontologiesSemantic Web: highly developed description

languages existMerger between web & ontologies envisaged Quantity of mark-up so far insufficient

Legal semantic webSemantic mark-up of legal information systems should

be re-used Field structures Thesauri Citations

AI & law (legal logic, conceptual information retrieval etc.)

Incorporation of world ontologies

16

Legal thesauriLegal thesauri

ISO 2788 standard Definition

Precompiled list of important words in a given domain of knowledge (controlled vocabulary)

Concepts are linked with relations Synonyms (polysems), antonyms, broader term, narrower

term, homonyms Dictionary: definitions

Information science + legal information systems Documentation and retrieval

Nucleus of a lexical ontology

17

Legal ontologiesExplicit formulation of a legal domain

Thesauri + definitions + more relations + formalisation for IT applications

Conceptual model Abstract, simplified, computable

New form of abstraction and formalisation of law Theory of formalisation (?)

Advantages Computable Links with world ontologies Re-use of existing ontologies Important tool for automation of law

Problems High efforts required for knowledge acquisition Scaling-up (well-known problem in AI & law)

18

Related workEarlier formalisation attempts

Hohfeld, Allen, McCarty, Stamper etc.1990ies

FOLaw (Valente), FBO (van Kralingen, Visser) Workshop on legal ontologies 1997

LOAIT Workshop on Legal Ontologies and Artificial Intelligence Techniques 2005 and 2007

ICAIL International Conference on Artificial Intelligence and Law Sessions on ontologies since 1997

LEX Legal XML Workshop Florence 2007Major research: Leibniz Center for Law, Amsterdam; ITTIG,

Florence; University of Turin, Autonomous University of Barcelona, University of Vienna etc.

19

Types of applicationsRepresentation of legal knowledge

e.g. FBO, LRI Core, LKIF

Conceptual information retrievalJuriservice, LOIS

Advanced lexical ontologiesMultilingual thesauri

e.g. LOIS, Legal Taxonomy Syllabus

Interchange of documents and knowledgee.g. MetaLex, eLaw

20

Knowledge representation (1)Language for Legal Discourse LLD / McCarty

(1989)NORMA / Stamper (1991)Frames-based ontology (FBO), van Kralingen and

Visser Common legal ontology; re-useable, 3 classes of model

primitives, for each class a frame structure has been defined with all relevant attributes

Functional ontology (FOLaw), ValenteAim: organisation and linking of legal knowledge, in

particular in respect to conceptual information retrieval6 basic categories of legal knowledge

Normative knowledge, meta-legal knowledge, world knowledge, responsibility knowledge, reactive knowledge, creative knowledge

21

Knowledge representation (2)ON-LINE (architecture of legal case-solving)PROSA (training system for legal case-solving)

E-Court, LRI-Core, University of AmsterdamGoal: semi-automated multi-lingual information

management for various sources (audio, video, text); application area: penal law

LRI-Core: broad concept structure with typical legal main concepts About 200 concepts, in development anchors Links between foundational (upper) ontology (=

world knowledge) and legal core ontology (legal concepts)

Supports legal subsumption 22

Knowledge representation (3)Select/direct from various acts or agents to the legally

relevant onesE-Power, project of the Dutch Tax and Customs

Administration Application-oriented knowledge system; formalisation of laws

and regulations as conceptual models Automated tasks (e.g. subsumption, calculation, document

assembly); comprehensive support from legislation to application

LKIF Core Ontology (Legal Knowledge Interchange Format) (Estrella project), University of Amsterdam Standard OWL ontology OWL-DL (description logic) Description logic programs (DLP)

23

Knowledge representation (4) Obligation, permission, roles, rights, duties, privileges,

liabilities etc. Top level clusters Mereological relations Location Time Changes (processes) Agents + actions + roles Propositions Legal agents + actions, rights, powers Norms LKIF rules – more expressive than OWL Application: traffic domain

Impressive standard

24

Multilingual thesauri (1)LOIS Lexical Ontologies for legal Information

ServingMulti-lingual access to European legal databasesFormal representation of legal concepts in all languages

on the basis of the WorldNet technology; similar concepts6 languages, 5000 synsets ILI inter-lingual index + legal definitions10 partners; leader: ITTIG, Florence

Legal Taxonomy Syllabus (University of Turin)Tool to annotate and recover multi-lingua legal

information (EU Directives)Legal dictionaries Taxanomies of legal concepts

26

Multilingual thesauri (2)DALOS (ITTIG, Florence)

Ontological-linguistic resource for multilingual drafting process (EU)

Basis: LOISOntological layer: conceptual modelling at a language-

independent level Lexical layer: lexical manifestations in different

languagesTerm extraction using NLP tools

27

Advanced lexical ontologiesLOISLegal Taxonomy Syllabus JuriserviceDALOS Comprehensive legal ontology (University of Vienna)

Real world (world knowledge) Legal system as a order of norms : socio-economic

governance by law with the goal of risk reduction Frames

Material rules Procedural rules

Concepts Concept frames

Starting point legal thesauri, e.g. LOIS thesauri Links: world knowledge, rules, top legal ontology

Hard core of a legal ontology

28

Interchange of documents and knowledgeInterchange standards for documents

Many international and European applicationse.g. EU, eLaw (Austria), MetaLex

Interchangability of legal knowledge representation MetaLex (University of Amsterdam)

Generic and extensible framework for XML-encoding of legal resources

29

Dynamic Electronic Legal Commentary (1)

Abstract representation of law in a conceptual & logical-systematic structure; like printed commentary but in a machine-useable format

Description of the world ([possible] facts)The core: links between possible facts (situations)

and legal consequencesProblem: world ontologies have still some way to

improve sufficiently, legal formalisation has to move from small environments to the real big world

It‘s time to move on

Dynamic Electronic Legal Commentary (2)For legal information systems: Not the very, very big step, but :

Tools like a navigator [time and document types, layers of the legal order, consolidated texts] (e.g. PreLex) , citator or terminologist are possible and would be highly desirable …

Thus good paying services

In the near futureThe real thing … some automated support for legal

subsumption, e.g. helping in the real game of applying legal provisions (could that also called legal reasoning or a legal expert system … maybe?)

European standard of legal ontologiesMotivators

Comparative legal research, harmonisation of EU law (e.g. Services Directive), European E-Government

Ontologies are standards, thus an obvious thing to doMeta information in Dublin Core Metadata

languageCitations (standard, URI)Ontological structures

RulesHaley Ltd. (formerly: Softlaw)

ConceptsE.g. lexical ontologies projects

32

33

Next steps (1) Interchange of documents and knowledge

Legal documents: ongoing and improving Legislative documents: many applications, standards like

MetaLex may improve formalisations due to inclusion of knowledge representation aspects

Improvement and enlargement of legal thesauri Up to 10.000 concepts

URI formalisation of citations on an EU level Multilingual information retrievalConceptual information retrieval using legal thesauri

Improved searching, classification and summarisation of documents

Word sense disambiguation for easier coupling with legal information systems

34

Next steps (2)Text analysis and text categorisation Start: information system (text archive)

Classification Concept analysis (e.g. DALOS, KONTERM) (Semi)automatic text analysis

Summaries (e.g. SALOMON, KONTERM, FLEXICON)Result: semantic description of the legal order; some

“primitive” anchors to legal system and world knowledge

Inclusion of results of text analysis in an advanced lexical ontology

35

Next steps (3)Concept frame

Header, definition, relations + More relations, better definitions, links to legal rules +

world knowledge “Raw” conceptual legal ontology

“Raw” dynamic electronic commentaryConceptual description of legal order with links to legal

rules and world knowledge Formalisation of dynamic electronic commentary

in LKIF or other ontology languages Big step, resources not availableMore research required

36

Conclusions Ontologies are the key for a computer-useable

formalisation of the knowledge on the world and the legal system

XML: standard for mark-up of legal documents XML/ontologies: emerging standard for knowledge

representation New form of a legal commentary: dynamic, electronic,

computer-useable Big support for European legal work

Legal search, exchange of data, exchange of knowledge Next steps

Exchange standards Multilingual information retrieval Improvement of legal thesauri Some (semi)automatic text analysis and categorisation,

advanced lexical ontologies Later: formalisation in LKIF

Big potential for easier better European legal work

37

Contacts

Erich Schweighofer

Universität WienArbeitsgruppe Rechtsinformatik

Wiener Zentrum für Rechtsinformatik

[email protected]

http://rechtsinformatik.www.univie.ac.at

IRIS2009 Internationales Rechtsinformatik Symposion, Salzburg http://www.univie.ac.at/RI/IRIS2009