Corporate Overview

1
Creating order from IT chaos?
Central Data eXchange (CDX) Then, Now and in the
Future Michael C. Daconta Chief Scientist,
Advanced Programs Group 8/28/2003
2250 Corporate Park Drive Suite 500
Herndon, VA 20171 703.326.1000
http//www.mcdonaldbradley.com
2
Agenda

• Introduction
• Then
• eMortgage experience
• VKB experience
• XML Training, Books Design experience.
• Now
• Unnecessary Complexity
• No integration strategy
• Limited Semantic Interoperability
• Future
• Streamline current system
• Federation
• Semantic Technologies
• Conclusion

3
Introduction

• Michael C. Daconta
• Director, Semantic Web Technologies Chief
  Scientist, Advanced Programs Group, MBI
• Chief DIA Architect, VKB Collateral Space/NCES
• Author/co-author of 10 technical books
• Inventor of Fannie Mae XML Electronic Mortgage
  Standards (2 patents pending)

4
Then Overview

• The proposal that almost was
• Lockheed Martin missed the CDX submission
  deadline by 20 minutes.
• Daconta as Key Personnel
• Would have participated in Orals
• Team leveraged MBIs architecture experience in
  proposal.
• Lessons Learned from
• Electronic mortgages XML development
• Teaching XML
• Virtual Knowledge Base Design and Implementation

5
Then eMortgage experience

• Chief architect of Fannie Maes eMortgage
  architecture.
• Adopted as a MISMO standard
• Patented by Fannie Mae
• Key achievements
• SMART Documents (patented)
• Rule-based validation (patented)
• Secure Transmission Package with Manifest

6
eMortgage Experience (2)

• SMART Document
• Separates presentation and data
• Arcs link/transform fields
• Lessons Learned
• Simple Structure
• No ID/IDREF
• Secured
• XML Signature
• Verifiable
• Validator
• Use of precedent and analogy in design

7
Then VKB Experience

• Virtual Knowledge Base
• Virtual
• Data Bus Architecture (web services)
• Standard Enterprise Services (producer/consumer
  federation)
• Knowledge Base
• Resource Registry

8
Then XML Design Experience

• XML since 1996
• Standard in 1998
• XML Training
• Web Services in 1999
• Standard in 2003
• Design articles
• Are Elements and Attributes Interchangeable?
• Other Design issues
• Containment versus IDREF
• Modularization
• Latest RPC versus Document-based Web Services
• Lesson XML Design matters

9
Now Overview

• Quick Review of key documents
• Blueprint for a National Environmental
  Information Exchange Network
• Core Reference Model
• XML Design Rules and Conventions
• XML Schemas
• Final Network node Functional Specification V1.0
• CDX Web Services Definition Language (WSDL)
• Following analysis of CDX and EIEN
• Caveat understanding based only on document
  review and conversations with Brand Niemann

10
Now Unnecessary Complexity

• Second-system pitfall
• Every technology goes through three stages
  first a crudely simple and quite unsatisfactory
  gadget second, an enormously complicated group
  of gadgets designed to overcome the shortcomings
  of the original and achieving thereby somewhat
  satisfactory performance through extremely
  complex compromise third, a final proper design
  there from. - Robert Heinlein, The Rolling
  Stones
• XML Design Complexities slow adoption
• Container approach simpler than ID/IDREF
• Always choose the simplest method to achieve a
  goal!
• XLink/XPointer not well supported/needed
• Multiple Namespaces is an advanced concept.
• Data Type Derivation must be carefully scoped

11
Now Unnecessary Complexity (2)

• Core Reference Model
• The metamodel reinvents the wheel
• Data Element, Data Block, Compound Data Block,
  Major Data Group ? better to stick with OOP
  terms.
• OOP adoption Java, .Net, UML, OWL
• Omits concept of Instances and Associations
• Too many parts need to Simplify and Streamline
• Precedent OSI versus TCP/IP
• Document-level and data-level distinction
  unnecessary
• Class is a recursive concept merge Data Block
  and Data Group (or better, switch to Classes).
• Simplify and Streamline with a clear vision (see
  future slides) will enable
• Aggressive completion of remaining formats ( lt 6
  mos.)
• Return to national initiative
• Universal exchange (or access) is a necessary
  pre-condition to higher level goals like semantic
  interoperability.

12
Now No Integration Strategy

• Hub and spokes architecture
• CDX is the exchange hub
• Point-to-Point XML exchanges
• No Information Integration
• Blueprint stated Information, especially
  integrated information, is an increasingly
  important environmental management tool.
• Whats the mechanism for integration?
• Virtual Knowledge Base federation concepts
• Resource Registration
• Data Producer Registration
• Data Consumer Registration
• DODs Net-centric Enterprise Services (NCES)
  Strategy
• Post to a shared, managed, network space
• Space accessible via portal interface or system
  interface

13
Now Limited Semantic Interoperability

• Core Reference Model standardization
• Some Shared Semantics
• Schemas Review
• Beach-Monitoring Data
• Good
• Good use of enumerations
• Bad
• Exclusively elements
• Weak hierarchy
• StartDate, StopDate ? concept of Duration
• BinaryLargeObjectTypeCode should use MIME type.

14
Schema Review (2)

• Drinking Water Bacteriological Analysis
• Good
• Robust hierarchy
• Bad
• Exclusively elements
• High tag to data ratio in example not enough
  optional or conditional tags?
• Certification statement should use XML signature
• Label redundancy i.e. MailingAddressCityName,
  MailingAddressText, MailingAddressStateName
• Mixed Metaphors SenderPhoneFaxEmail versus
  SamplerContactInfo ? both are Contact
• Facility versus PWSFacility ambiguity.

15
Schema Review (3)

• National Emissions Inventory Schema
• Only reviewed Area Non Road Submission Group
• Good
• Good Type creation/constraints
• Bad
• Exclusively elements affects performance since
  sample file is 2.6 MB. Poor design decision.
• Uses different elements for concepts Contact
  and Duration in other schemas.

16
Schema Review (4)

• Facility Registration System V2.2
• Good
• Good documentation
• Good modularization
• Good hierarchy
• Bad
• Could use better semantics
• Person versus Individual Data Type
• individual is a broader concept
• FacilitySiteTypeName is a string?
• Most types should be enumerations.

17
Schema Review (5)

• CDX WSDL
• Soap encoding violates WS-I Basic Profile
• Paragraph 5.6.4 ( see http//www.ws-i.org/Profiles
  /Basic/2003-08/BasicProfile-1.0a.htm)
• Should use Document-based web services
• (See http//webservices.devchannel.org/webservices
  /03/07/11/2122220.shtml?tid25tid38)
• Type should use MIME type.
• Document structure should include Dublin core
  metadata.
• Only binary content should be base64 encoded
  (expands size by 1/3).
• Operations are too generic to serve as federated
  web services. (See IFIS in Future slides)

18
Future Overview

• Initial Recommendations
• Streamline, Simplify and complete Current System
• Establish a distributed Federation
• Leverage Semantic Technologies
• Each recommendation needs further study and
  refinement
• And aggressive Implementation!

19
Future Streamline Current System

• Simplify
• Remove complexities from guidance
• Streamline
• Remove unnecessary parts
• Leverage Enterprise Schema Management Tools
• SchemaLogic, BlueOxide, etc.
• Complete
• Aggressive drive to 1.0 drafts for all data
  exchanges
• Most efficient mechanism expert draft followed
  up by a workshop with the influenced
  organizations to hammer it into a 1.0.

20
Future Federation

• What is a federation?
• A group of cooperating independent entities.
• Also known as grids, fabrics, service-oriented-arc
  hitectures and distributed systems.
• Examples
• Digital library federation, grid-computing, jini,
  agent federations, ebXML and UDDI federated
  registries
• Requirements for Data federation
• Registry
• Web-service polymorphism (static or dynamic
  binding)
• Reliability and failover

21
Future Federation (2)

• NCES Federation Prototype (OSD program)
• Intelligent Federated Index Search (IFIS)
  currently integrates 6 enterprise systems.

22
Future Semantic Technologies

• Overview
• Enterprise Architect Article
• Smart Data Continuum
• NCES impact put the smarts in the data and NOT
  the application endpoints
• Semantic Technologies for eGov Workshop
• Purpose of Semantic Web Book
• Understandable by non-PhDs
• Show the Return On Investment (ROI)
• Provide a Roadmap

23
Future Semantic Technologies (2)

• Data Independence is step one.
• Data is more important than applications.
• Data value increases with the number of
  connections it shares.
• Data about data can expand to as many layers as
  there are meanings.
• Data modeling harmony is the alignment of syntax,
  semantics, and pragmatics.
• Data and logic are the yin and yang of
  information processing.
• Data modeling makes the implicit explicit and the
  transparent apparent.
• Data standardization is not amenable to
  competition.
• Data modeling must be decentralized.
• Data relations must not be based on probability
  or luck.
• Data is truly independent when the next
  generation need not reinvent it.

24
Conclusion

• Then
• Experience matters.
• Now
• You are on the right track (though progress has
  been slow)
• Current problems are fixable.
• Need aggressive champions with expertise to
  finish!
• Future
• Get to federation.
• Get to Semantic Interoperability.
• McDonald Bradley, Inc. can help you get there!
• Questions?