Growing the Tree of Agreement:

1
Growing the Tree of Agreement

• Building, Managing and Disseminating Metadata
  Standards Through Online Collaboration

Online Northwest 2004 - February 20, 2004
Presenter Breanna Anderson CTO SchemaLogic
Inc. breanna_at_schemalogic.com http//www.schemalog
ic.com
2
Metadata Matters

• "Companies are beginning to experience taxonomy
  creep, as they deploy multiple content management
  systems across the enterprise and the frequency
  of taxonomy updates required by these distributed
  systems accelerates. Many companies are growing
  beyond the capabilities of existing static
  taxonomy programs and, as a result, deployment of
  automated taxonomy management solutions will
  increase through 2005/6.
• Tim Hickernell, META Group

3
Metadata Matters

• "The greatest issue facing organizations today
  with regards to information management is
  heterogeneity. Information comes in many
  different formats and stored in many different
  systems.  Each has its own respectively correct
  schema.  Information consumption happens for many
  different business reasons and by many different
  user communities, from RD staff to customers.
  Each has its own needs and lexicon.  The solution
  required is one of reconciliation, transparently
  and seamlessly bridging the knowledge across
  systems and user communities in a controlled and
  centralized manner.
• Carl Frappaolo, Delphi Group

4
Turtles, Turtles
5
A Balanced Approach to Schema Standards
Managed Diversity
The One Model
NM Map
6
The 8-Fold Path to Meta Data Enlightenment
A workable metadata modeling and management
solution must be

• Complete
• General
• Granular
• Consensus Managed
• Culturally Adaptable
• Consumer Oriented
• Actionable
• Evolutionary

7
8-fold Path RevealedComplete

• Must address Taxonomic metadata standards in a
  flexible way. Provide powerful tools for
  Vocabulary/Thesaurus creation, rationalization
  and management.
• Must provide for structural schema modeling in a
  way that encourages consistency but allows
  diversity.
• Must reflect the interactions between taxonomy
  and structure with integrated change management.

8
8-fold Path RevealedGeneral

• Should not be biased unduly toward one specific
  implementation or encoding of the schema.
• (XML Schema does not work well for Relational
  systems and SQL DDL does not apply well to XML
  etc.)

9
8-fold Path RevealedGranular

• Schema Definition objects should be managed as
  proper objects with lifetime identifiers, highly
  granular reusability, their own metadata,
  permission control and auditability.
• Managing definitions in arbitrary file-based
  document groupings inhibits reusability,
  governance and manageability

10
8-fold Path RevealedConsensus Managed

• When Definitional objects become group property
  and when reuse and factoring occur, automatic
  impact analysis and consensus management is
  necessary to address the Fragile Superclass
  problem.
• Participation in the collective requires security
  for individual interests.

11
8-fold Path RevealedCulturally Adaptable

• Not all organizations work in the same way. The
  Schema Management solution will impact the
  culture but must also reflect, respect and adapt
  to the organizational culture.
• Axes Top down mandates/Bottom Up consensus Fast
  Paced/ Deliberate, Meeting-oriented/Spec-Oriented.
  
• Note Different parts of an organization may have
  very different sub-cultures.

12
8-fold Path RevealedConsumer Oriented

• Must be usable, Directly, by Knowledge Domain
  Experts People who know the details of the
  problem to be solved, but are not necessarily
  Metadata Wonks.
• Should support management of meta-schema about
  how End-Users will see/use the Metadata
  Localization, Display Tips etc. Not just
  information for machines to interpret because
  People are the ultimate (meta)data consumers.

13
8-fold Path RevealedActionable

• Should allow or actively support automated
  propagation and implementation of Metadata/Schema
  changes into relevant systems.
• This implies an adequate level of specific
  detail, possible application-specific information
  while not violating the Generality principle.
• Requires registration of external dependencies on
  managed schema objects, meta-model transform
  rules and possible handling of data impacts.

14
8-fold Path RevealedEvolutionary

• Should be built to account for the fact that
  standards change over time and that multiple
  variants of a given metadata definition may
  coexist over time as a family of related
  standards.
• External systems of record must be accommodated
  by all but the most authoritative metadata
  management system.

15
Defining Terms

• There is a singular lack of vocabulary control
  in the field of controlled vocabularies, - Bella
  Hass Weinberg, professor of library science, St.
  John's University, New York

16
The Great Thing About Standards

• Encoding SGML, EDI, XML, Flat File
  Metadata/Schema DTD, XML Schema (SXD), XDR,
  RELAX NG, RDF, ISO 11179,UML,EDI, MOF, OWL, XMI
  (XML Metadata Interchange)
• Taxonomy Standards ANSI/NISO z39.19, OWL, RDF,
  Topic Maps, ISO 11179
• Service Definition/Execution WSDL (SOAP),
  CORBA, Java-RMI, DCOM

17
The Great Thing About Standards Continued

• Directory /Discovery UDDI (SOAP), LDAP, MS
  Active Directory, Novell Directory Service
• Metadata/Schema Standards
• Dublin Core
• DISA XML Emporium
• METS Metadata Encoding and Transmission
  Standard
• MPEG-7 image metadata standards
• RSS Really Simple Syndication
• PRISM www.prismstandard.org
• XMP Adobe, Extensible Metadata Platform
• ICE - Information and Content Exchange
• eBXML
• OASIS Umbrella for multiple efforts

18
Practical Suggestions

• Make it easier to reuse than to reinvent
• Expose the costs of unilateralism
• Underscore the value of collectivism
• Facilitate local optimization where necessary in
  an extensive fashion (avoid total opt-outs)

19
Practical Suggestions

• Metadata/Schema/Taxonomy standards development or
  rationalization projects must have a compelling,
  demonstrable goal to be successful long-term.
• Keep scope modest initially until payback of time
  and investment is established. Dont boil
  the ocean by making the scope to broad or deep.

20
Practical Suggestions

• The tree of agreement presumes a consistent
  foundation that allows for variations in ultimate
  usage of meta-models
• The Ultimate modeling system should be singular,
  flexible and consistent and should be applicable
  to multiple outputs.
• Otherwise, modelers will spend all of their time
  mapping terminology and semantics during the
  modeling process and not address the underlying
  conceptual problem

21
Practical Suggestions

• Secure high-level sponsorship before proceeding
  too far to avoid wasting everyones time and
  energy.
• Proving ROI can be difficult at this point.
  Stick to fundamentals of good management and
  governance and ensuring success of existing
  projects and investments.

22
An Example Active Metadata Repository

• SchemaServer as an example of an active
  Schema/Metadata repository for modeling and
  collaboration.

23
(No Transcript)
24
Content Class with Associated Elements
Content Classes are comprised of Elements to
define your Structural Metadata requirements
Element Properties can be customized per Content
Class
25
Modeling Overlap and Diversity with Inheritance
26
Inheriting Elements and Extending your Metadata
model
Within your Metadata Model, the Semantic Tree of
Agreement provides inheritance and extensibility
to your Metadata model
27
Element Properties define the concept
Elements define the actual descriptions of
data Vocabularies can be associated with elements
to define the list of allowed values for the
concept
28
Vocabularies

• Terms are entities unto themselves may live in
  multiple vocabularies
• Have their own properties
• May be presented and subset in many different
  ways
• Rich thesaurus modeling capabilities
• May be localized

29
Metadata Reuse-XML Schema output
30
Metadata Reuse-Web Form
31
Metadata Reuse-Database
32
Collaborative Change Control
33
Collaborative Change Control

• Contract - Driven Change Control
• Email Notification
• Impact Analysis shows who and what is affected by
  potential changes to shared objects. Approval
  process ensures awareness and consensus prior to
  a change being made.
• Role-based permission management lends itself to
  data stewardship processes by ensuring ownership
  and accountability.

34
Change Management with Online Voting
35
My Work Area Shows all proposed changes
initiated by the user and all changes the user
needs to vote on.
36
(No Transcript)
37
SchemaServer 2.2 Detail Change Management Workflow
38
SchemaServer Availability

• SchemaServer available for attendees of Online
  Northwest Conference 04
• www.schemalogic.net/olnw
• Email breanna_at_schemalogic.com for username and
  password access