1 of 29

1
XML as a Strategic Information Technology for DOE

• Frank Olken John L. McCarthy
• Lawrence Berkeley National Lab
• InterLab 99
• Stanford Linear Accelerator Center
• November 4, 1999

2
Summary Overview

• XML can help DOE share information
• Why XML?
• What DOE applications could benefit?
• Requires additional standards/infrastructure
• content standards
• code sets
• repositories / registries

3
LBNL Metadata Activities sponsored by US EPA
Superfund

• W3C RDF, XML Schema, XML Query Lang. WGs
• ISO SC32 WG2 / NCITS L8
• ISO 11179, ANSI X3.285, XMI
• XML/XSL Prototype for metadata exchange
• OMG - XMI, CWDMI
• Ecommerce - eCo Framework, OASIS/CEFACT

4
Where can (should) DOE use XML?

• EDI / Ecommerce
• Scientific Engineering Data Exchange
• Remote Procedure Call / Object Serialization
• Scientific Data Archiving
• Publishing (scientific, other)
• Metadata Exchange
• Information Discovery, Query, Integration

5
XML provides syntax, but data sharing also
requires other standards layers

• Structure Standards (e.g., XML Schema)
• Content Standards (e.g., geospatial, chemical,
  ...)
• Repository Standards (e.g., ISO 11179)
• Measurement units standards
• Digital Signature Infrastructure
• Repositories of standard data elements, code
  sets,
• .

6
Analogy to Human Language

• Communication requires shared language
• Shared alphabet
• Shared vocabulary
• Shared syntax and structure
• We spend years teaching children language/reading
• We spend more years teaching specialized
  vocabularies (medicine, chemistry, .)

7
Attractions of XML

• XML is essentially an LL(1) parser generator
• Human readable, editable, ...
• Little semantics specified
• Applies to both documents and data
• Can encode many different data models
• Can encode programming languages (e.g. for
  workflow)
• Many tools are becoming available
  (parsers , formatting engines, repositories,
  query tools, )

8
Potential XML Competitors?

• ASN.1 - data only (no docs), often not human
  readable, richer base types, smaller tool base
• netCDF, HDF5 - binary, data only, not human
  readable
• SGML - limited tool base (no browsers)
• X12 EDI messages - positional, idiosyncratic
• comma delimited files - positional, error prone
• STEP Part 21 files - product data, narrow tool
  base
• domain specific formats - ad hoc, small tool base

9
XML Data Model

• Eclectic (SGML Legacy)
• Hybrid hierarchical / network
• hierarchical tree of nested elements
• network IDREFs, XLINKS
• Often described via graph data model
• Can encode OO, relational other models
• but may need to query higher level data models

10
XML for DOE EDI / Ecommerce

• CME (FTPA proposal database)
• But inconsistent budget account structures, etc?
• Procurement
• Travel
• Human resources (job postings, resumes)
• Financial reporting to DOE
• Enabling technology for workflow management

11
XML for Scientific Data Exchange

• Applications
• Astronomy
• Bioseqence ML
• ChemML (chemical compound markup)
• XML is verbose - unsuited for very large
  databases
• XML Schema language currently lacks vectors,
  arrays, complex numbers, measurement units, ...

12
XML for Scientific Data Archiving

• Similar requirements to scientific data exchange
• Preserve data and meaning over time
• Requires preservation of data and programs to
  decode / access data
• Attractions of XML - human readable, long lived
  standard, many tools
• e.g., San Diego Xarchive
• LLNL ?, LANL ?

13
XML for Engineering Data Exchange

• Mechanical / architectural CAD data exchange
• XML encoded STEP Part 21 files Part 22
• aecXML (for architecture / construction)
• Major application arena for DOE
• EPRI / NERC CCAPI - power systems config.

14
XML for Scientific Publishing

• Single source document for multiple media outputs
• Much more flexible powerful than HTML for print
• MathML for math, ChemML for chemistry
• MathML, ChemML tools recently available

15
XML for RPC / Object Serialization

• RPC remote procedure call
• used in place of XDR, IIOP,
• used for Java object serialization (comm/storage)
• more verbose, human readable
• many possible transport protocols (HTTP, email,)
• examples XML RPC, Web Methods WIDL

16
XML for Metadata Exchange

• XMI - UML model interchange (DB schemas, )
• RDF (Resource Description Framework)
• Content Standards (Dublin Core, GILS, FGDC, )

17
XML for Information Discovery, Query
Integration

• Semantic markup permits more precise retrieval
• Content standards Dublin core, FGDC, MARC
• Energy abstracts, molecular biology annotations /
  abstracts
• XML Query Language being standardized
• Use of XML for mediators, data integration
• SDSC, Stanford, ATT, U of Washington,
• XML can transform WWW to World Wide Database

18
XML Structural Standards Needed

• XML (done)
• XML Schema Language (1Q 2000)
• XML Query Language (4Q 2000?)
• XMI (V1.0 done)
• Process (workflow) Protocol modeling
• Petri Nets
• Coupled Finite State Automata

19
XML Content Standards Needed

• XML Schemas
• Domain Data Models (schemas)
• e.g., workflow, ...
• Message Content Specifications
• views over domain data models
• XML/EDI, HL7
• Standard code sets, vocabularies
• Accounting Standards (e.g., overhead definitions)

20
Standard code sets, vocabularies

• Element / Subatomic particle names/symbols
• Chemical names
• Geographic place names, codes
• SNOMED (std. nomenclature for medicine)
• LOINC (std. clinical lab test names, )
• X12 message codes

21
Measurement Units Stds. Needed

• Conventions for specification of
• Measurement units (meters, feet, kilograms)
• Dimensionality (length, mass, time)
• Automatic inference/checking of
  dimensionality/units consistency in query
  expressions

22
Repositories Needed

• Thesauri, ontologies data element registries,
• Means of sharing code sets, vocabularies
• Means of sharing schemas, message formats
• Implies shared metamodel
• Requires access interface (APIs, query language)

23
Data Element Registry Activities

• ISO 11179, ANSI X3.285
• EPA Environmental Data Registry
• HCFA
• DOD Healthcare
• OASIS (Ecommerce)
• Australian Healthcare

24
Digital Signatures Needed

• Essential if digital documents are to replace
  paper
• Need digital signature standards (several exist)
• W3C Digital Signature Working Group
• Need certificate authorities (coming slowly)
• May need smart cards (slow in US)
• Need legal acceptance (most states, pending in
  Congress)

25
Current DOE Standards Infrastructure Efforts

• Data exchange formats HDF5 for ASCI DMF
• Data models Fiber bundle for ASCI DMF
• Infrastructure Development
• digital signature certificate authorities
• Repositories ???
• Standards development
• DOE is AWOL

26
You are invited to participate in the Open Forum
on Metadata Registries

• January 17-21, 2000
• Santa Fe, New Mexico
• Sponsored by ISO SC32 WG2, JTC1, EPA
• Presentations of ISO11179, X3.285, XMI,
  implementations,
• Disciplinary tracks environmental, healthcare,
  ...
• http//www.sdct.itl.nist.gov/ftp/l8/sc32wg2/2000/
  events/openforum/index.htm

27
Conclusions

• XML could become strategic IT for DOE
• Facilitates information sharing
• Needs complementary standards
• Requires more infrastructure
• DOE needs to get its act together .

28
Acknowledgements

• Paid for by U.S. Environmental Protection Agency
  Superfund Office
• No DOE funding yet .

29
Contact Information

• Frank Olken, LBNL
• Mailstop 50B-3238
• tel. [phone]
• olken_at_lbl.gov
• http//www.lbl.gov/olken
• John McCarthy, LBNL
• Mailstop 50C
• tel. [phone]
• jlmccarthy_at_lbl.gov
• http//www.lbl.gov/mccarthy