Akoma Ntoso as a standard for the lifecycle and the transparency of legal and legislative documents

1
Akoma Ntoso as a standard for the lifecycle and
the transparency of legal and legislative
documents

• 10th International "Law via the Internet"
  Conference, Durban, South Africa
• 26 - 27 November 2009
• prof. Fabio Vitali
• Department of Computer Science
• University of Bologna prof. Monica Palmirani
• CIRSFID Interdepartmental Centre of ICT Law

2
Summary

• Akoma Ntoso
• The main structure of the document
• The deep structure of the document
• Basic metadata
• Advanced metadata
• Applications of metadata
• Conclusions benefits of the adoption

3
AKOMA NTOSO

• It is an open legal XML standard for
  parliamentary, legislative and judiciary
  documents
• Promoted by the UNITED NATIONS Department for
  Economics and Social Affairs (UN/DESA) in 2004
  from the Kenya Unit
• It means Linked Hearts a symbol used by the
  Akan people of West Africa to represent
  understanding and agreement but it is now
  promoted also in Latin America, Asia and European
  regions

4
AKOMA NTOSO

• Architecture for Knowledge-Oriented Management of
  Any Normative Text using Open Standards and
  Ontologies
• Describes structures for legal documents in XML
• References documents across countries using a
  common naming convention - URIs
• Adds systematic metadata to documents using
  ontologically sound approaches
• Aims to
• Be extensible for the individual needs of any
  country
• Preserve the legal digital resources over time
• Guarantee legal principles
• Favour trust (authoritative versions, legal
  copies, etc.)

5
Learning to swim the structure of documents
6
Managing the structure of the document - 1

• An Akoma Ntoso document is either an act or a
  bill (legislative documents), or a report or a
  debateRecord (debate documents) or a judgment, or
  a generic document.

• All Akoma Ntoso documents start with a metadata
  section, followed by an initial part (e.g., a
  preface, cover page, preamble, etc.) followed by
  the body of the document, and then a conclusion
  and possibly one or more attachments.

lt?xml version"1.0" encoding"utf-8"?gtltakomaNtoso
gt ltact contains"originalVersion"gt ltmetagt
lt/metagt ltprefacegt lt/prefacegt ltpreamblegt
lt/preamblegt ltbodygt . lt/bodygt
ltconclusionsgt lt/conclusionsgt
lt/actgt lt/akomaNtosogt
7
Managing the structure of the document - 2

• Within the main body of the document, each
  document type has its own structure.

• A hierarchy of parts for legislation
• E.g. section, part, paragraph,chapter, title,
  book, tome, article, clause, etc.

ltpreamble id"preamble"gt ltpgtAn Act of
Parliament to amend the Retirement Benefits Act,
1997 lteol /gtENACTED by the Parliament
of Kenya, as follows -lt/pgt lt/preamblegt
ltbodygt ltsection id"sec1"gt
ltnumgt1.lt/numgt ltheadinggtShort
title.lt/headinggt ltclause id"art1-cla1"gt
ltcontentgt ltpgtThis Act may
be cited as the Retirement Benefits (Amendment)
Act, 2003.lt/pgt lt/contentgt
lt/clausegt lt/sectiongt

• Any of a list named sections for debates
• E.g. questions, answers, notices of motions,
  procedural motions, etc.

• A sequence of named sections for judgements
• Introduction, background, motivation, decision

8
Swimming in the pool the semantics of content
fragments
9
Managing the semantics of the text

• Text fragments describing and contextualizing the
  documents
• In all documents docType, docTitle, docNumber,
  etc.
• In judgements, also courtType, neutralCitation,
  party, judge, etc.
• References
• Definitions
• References (individual, multiple, ranges)
• Quotations (individual, multiple, ranges)
• Mention of relevant concepts
• Times, dates, relevant entities
• More later.

ltsection id"art2"gt ltnumgt2.lt/numgt
ltheadinggtAmendment of ltref href"/ke/act/1997-08-
22/3/eng/mainart2"gtsection 2 of No 3
of 1997lt/refgtlt/headinggt ltclause
id"art2-cla1"gt ltcontentgt
ltpgtThe Retirement Benefits Act, 1997, is amended
-lt/pgt ltlist id"art2-cla1-lst1"gt
ltitem id"art2-cla1-itma"gtltnumgt(a)lt/numgt
ltpgtby deleting the definition of
"financial year" and ltmod id"mod6"gtltref
id"ref2" href"/ke/act/1997-0
8-22/3/eng/main"gtsubstitutinglt/refgt therefore the
following new definition -
"ltquotedText id"mod6-qtd1"gtfinancial year" -lteol
/gt (a) in relation to the
Authority, has the meaning assigned to it in
section 19'lteol /gt (b) in
relation to a scheme, means such accounting
period as may be prescribed
in the scheme ruleslt/quotedTextgtlt/modgtlt/pgt
10
Swimming organizing the content in XML

• So far, we have organized in XML a legal document
  identifying its parts and providing a semantic
  description of the main structure and the most
  relevant inline fragments, including references
  and quotations.
• This is enough to provide for
• Display on screen
• Print on paper
• Hypertextual links
• It can be taken care of by a lower secretary in a
  back office with limited knowledge of legal
  documents, no knowledge of XML and a modified
  text editor (e.g. Bungeni).

11
Swimming organizing the content in XML

• So far, we have organized in XML a legal document
  identifying its parts and providing a semantic
  description of the main structure and the most
  relevant inline fragments, including references
  and quotations.
• The basic structures of Akoma Ntoso provide
  support for
• Different law systems (civil law, common law)
• Different legal traditions
• Descriptive contexts (in which the markup can
  only observe whatever structure was used in the
  document - legacy documents)
• Prescriptive contexts (in which the markup can be
  used to force desired structures and require the
  presence or abcence of some elements).
• The only requirements are
• Blind obedience, i.e., strict adherence to the
  wording of the document (a tome is a tome if it
  is called a tome)
• No lie, i.e., no confusion between content and
  interpretation (e.g., missing document
  information that should be found in the preface
  but aren't, are added in the metadata section,
  and not in the preface).

12
Swimming organizing the content in XML

• The main structural and semantic elements for the
  content of legislative and legal documents in
  Akoma Ntoso are enough to provide for
• Display on screen
• Print on paper
• Hypertextual links
• The identification of the right constructs for
  the organization of the document can be taken
  care of by a lower secretary in a back office
  with limited knowledge of legal documents, no
  knowledge of XML and a modified text editor (e.g.
  Bungeni).
• This is appropriate with the settings of many
  legal drafting offices around the world.

13
First dives basic metadata
14
Basic metadata

• The structure of metadata in Akoma Ntoso is
  complex but can be studied piecemeal.
• Publication, keywords and notes are easy to deal
  with.
• They require some higher grasp of legal aspects,
  and probably cannot be drafted by a lower
  secretary.
• They help in contextualizing the document and
  searching for it according to themes and
  theasuri.

ltpublication date"2003-09-04" name"Government
Gazette 25437" showAs"Government Gazette
25437" /gt
15
Scuba diving FRBR
16
FRBR the problems

• We would like a language that allows references
• To be either static or dynamic, according to need
  and legal nature of the text and of the reference
  itself
• And thus when crossing a hypertext links brings
  the reader to the right version of the right
  document
• To be independent of the technological choices of
  the repository of the documents
• And thus allows documents containing references
  to be moved to different machines, different
  server organizations, different server
  technologies, etc. guaranteeing the survival of
  the document collection in time.
• Idea the problem is NOT in the reference, but in
  the concept of documents itself.
• The Functional Requirements for Bibliographic
  Records (FRBR), a conceptual model for
  bibliographic items by the International
  Federation of Library Associations (IFLA),
  provides an answer

17
The FRBR model

• Every document has four different aspects
• The Work represents the abstract concept of a
  document, across all its versions, languages,
  representations. The South African Act 12 of
  2005 is a Work.
• Each Work is realized in one or many Expressions,
  which are concrete selection of textual content.
  Each expression defines a specific version of the
  content, in time and language. The English
  version as of the 1/12/2009 of the South African
  Act 12 of 2005 is an Expression.
• Each Expression is embodied in one or more
  Manifestations, actual representations of an
  Expression. Each manifestation chooses a computer
  format with its set of metadata for the
  expression. The Akoma Ntoso 1.0 version, with
  metadata by John Smith, of the English version as
  of the 1/12/2009 of the South African Act 12 of
  2005 is a Manifestation.
• Each manifestation is exemplified in one or more
  Items, physical copies of a Manifestation. Each
  item is a specific file stored on a computer. The
  copy on my computer of the Akoma Ntoso 1.0
  version, with metadata created by John Smith, of
  the English version as of the 1/12/2009 of the
  South African Act 12 of 2005 is an Item.

W
E
E
M
M
I
I
18
Using FRBR

• Each level of FRBR has its own address (URI).
• Higher levels refer to abstract concepts, and are
  used in document. The item level is a physical
  URL and is never used in document, but only in
  resolving an abstract reference.
• Thus changes in the physical organization and
  technology of the document repository does not
  require changes to the documents
• Dynamic references are references to works,
  static references are references to expressions
• When the document changes, a new expression
  exists, and the resolver will identify the
  expression that is most appropriate to a work
  reference
• Even when the document change, a static reference
  needs to be able to point to the old version of
  the document. Expressions never change, but are
  only added.
• Addresses (URI) of FRBR levels are similar
• URI of lower levels of the FRBR chain are a
  composition of the higher level addresses
  metadata specific of the level.
• It is easy, given an address of a Manifestation,
  to identify its Expression, and viceversa.

19
Things we get for free from FRBR

• Automatic support for multilinguism
• If legislation exists in multiple language, links
  allow you to traverse references always in the
  same language without being asked.
• Multiple repositories of the same documents
• Some authoritative, some not. Some complete, some
  selected. Some commented, some not.
• Non-authoritative consolidation of texts
• Especially useful in those countries (e.g.,
  Italy) where only a selected few acts are
  authoritatively consolidated.
• Point-in-time versions and change tracking are
  immediate derived functionalities of consolidated
  texts.
• Multiple metadata and comments of the same
  documents
• Different scholars and editors could add
  different sets of metadata elements and provide
  different views of the document
• Different selections of content
• E.g., private publishers could be interested in
  printing only a relevant fragment fo the act,
  omitting the rest (element ltomissis/gt )

20
Deeper and deeper TLC
21
Top Level Classes (TLC)

• Top Level Classes is the Akoma Ntoso mechanism to
  provide unambiguous references to concepts,
  roles, organizations, individuals.
• A formal conceptualization (technically, an
  ontology) has been realized for the concepts that
  are relevant to legislative and legal documents.
• It is composed of 10 independent classes (top
  level)
• Classes can be subclassed at will (e.g. Kenyan MP
  are the subclass of TLCPerson whose nationality
  is Kenyan and whose role is MP).
• Each individual is associated to a unique URI
  across time and documents (e.g., the same MP
  appearing in different parliamentary hansards may
  be shown with a different spelling for the name,
  but will have the same URI)
• Each reference in the document to a precise
  concept, individual, organization, role, is
  marked up with an ltentitygt eleemnt referring to a
  TLC instance in the ltreferencesgt section.

ltreferences source"FV"gt ltTLCRole
href"/ontology/role/political/MES" id"MES"
showAs"Minister for
Education and Sports" /gt ltTLCPerson
href"/ontology/person/ken/MP/gha.John.Gidisu"
id"per07"
shortForm"Mr. J.K. Gidisu" showAs"Mr. Joe
Kwashie Gidisu"/gt ltTLCPerson
href"/ontology/person/ken/MP/gha.John.OsafoMaafo"
id"per08"
shortForm"Mr. Osafo-Maafo" showAs"Mr. Yaw
Osafo-Maafo"/gt lt/referencesgt ltdebategt
ltquestion by"per07" to"MES"gt ltfromgtMr.
J.K. Gidisult/fromgt ltpgtasked the Minister
for Education and Sports the organic relations
between lt/pgt lt/questiongt ltanswer
by"per08" as"MES" gt ltfromgtMr.
Osafo-Maafolt/fromgt ltpgtMr. Speaker,
lt/p lt/answergt
22
Advantages of Top Level Classes

• We can now identify concretely persons,
  organizations, roles, concpets, places, across
  documents, spellings, languages.
• Meaningful searches across documents (e.g., all
  speeches given by the Minister of Finance in
  2009, or all acts mentioning company X) can be
  looked up with one simple query.
• The actual position of the reference is also
  easily found (a problem if there are differences
  of spellings and a long text).
• We have a solid foundation for more sophisticated
  inferences by exploiting the ontological
  framework underneath
• E.g., give me all documents in 2009 that contain
  a benefit (a TLCconcept) for company X (a
  TLCOrganization), or any company owned by company
  X or recursively owned by a company that is owned
  by company X.

23
In the deep sea consolidation, workflow and
lifecycle
24
Consolidation, workflow and lifecycle

• First assumption each document is the output of
  a specific step of a workflow and may change over
  time.
• each intermediate output is an FRBR expression of
  the same FRBR work. We just need to associate
  each expression to a specific step in a workflow.
  
• E.g. bill draft as proposed by MPs, as approved
  after first reading, as approved after second
  reading, as ready to be promulgated as act.
• Second assumption each document undergoes
  modifications due to events characterizing its
  lifecycle. Each event is the product of a
  specific document containing relevant information
  for the lifecycle, including modifications in
  validity, efficacy, and content.
• Each event that changes content creates a new
  FRBR Expression with the modified content.
• If the new content is not authoritatively
  produced, then it is possible to automatically
  consolidate the content by applying all relevant
  modifications.

ltlifecycle source"cirfid"gt ltevent id"e1"
date"1997-08-22" source"ro1" type"generation"
/gt ltevent id"e2" date"2003-12-19"
source"am1" type"amendment" /gtlt/lifecyclegtltre
ferences source"cirfid"gt ltoriginal id"ra1"
href"/ke/act/1997-08-22/3/eng/main"
showAs"Retirement Benefits Act" /gt
ltpassiveRef id"am1" href"/ke/act/2003-12-10/8/en
g/main" showAs"Amending Act" /gtlt/referencesgt
25
Classification in judgments
26
Conclusions Benefits and fishes
27
Benefits of the adoption of Akoma Ntoso

• An Akoma Ntoso repository can mark up the text to
  the level it feels appropriate.
• It is not necessary to understand, even less
  adopt, the more complex parts of the language.
• Basic document structure, references to other
  documents and basic metadata are enough for most
  repositories, both authoritative and not.
• But if the need arises, the tools are there.
• it is not necessary to adopt a different standard
  and convert every document,but one can simply add
  the new information.
• Even, added metadata can be provided in a
  separate document.
• Scholars, special interest groups, political or
  economical organization can provide any missing
  information and metadata on top of the
  authoritatively produced Akoma Ntoso documents
  for their readers and constituency.
• Since the underlying format is the same, the
  presence of new information is straightforward,
  smooth, transparent to the user.

28
BungeniEditor- open source Open Office markup
editor
29
References

• www.akomantoso.org
• www.parliaments.info, info at info_at_parliaments.inf
  o
• BungeniEditor on googlecode forum
• thank you for your attention
• Fabio Vitali fabio_at_cs.unibo.it
• Monica Palmirani monica.palmirani_at_unibo.it