XML Community of Practice

Meeting Notes

March 16, 2005

The focus of this meeting was XML registry services, with the Department of Commerce (DOC) hosting. Attendees introduced themselves and Owen Ambur noted that the prototype for the ET.gov site will be unveiled at GovCon05 on March 30. He also announced that Microsoft has scheduled a town hall meeting on InfoPath for government agencies on April 7 at FOSE.

Introducing Holly Hyland of the Federal Student Aid program, Owen said FSA’s XML registry has by far the best user interface of any he has seen thus far. Holly outlined the history of FSA’s involvement with XML and said they produced their first XSD a month after the W3C released the XML recommendation. They developed a student-centric XML data packet for financial aid and made it extensible to accommodate incorporation of other programs later. In response to requests, they extended the scope of the registry to encompass higher education generally. Specifying a standard for E-mail proved to be the biggest point of contention.

When FSA’s new CEO came on board and saw the agency’s modernization blueprint, he asked “where’s the data?” He directed the development of a data architecture and metadata repository for FSA, and the XML registry/repository is viewed as a collaborative tool for the data architecture. Accenture developed the registry under contract to FSA at a cost of $290,000. Monthly operational costs are only $1.98, which is donated, meaning the registry operates at no ongoing cost to FSA. It runs on Java, WebSphere and Oracle 9i and conforms to the ebXML 2.5 specification. XML is automatically generated based upon metadata in the registry.

Prior to beginning his overview presentation on ebXML 3.0, Joe Chiusano observed that FSA’s registry is generally compliant with ISO 11179, whereas ebXML is not. Also, Joe said that the term “ebXML” is a bit of a misnomer as far as the ebXML registry specification is concerned, because the specification is more abstract and does not apply directly to XML per se, since any kind of object can be managed in an ebXML registry. Sun Microsystems, for example, is using the registry for general purpose content management. In addition, with respect to reference to registry “object,” Joe suggested the term “entry” better describes the concept since a registry object is akin to an entry in a library card catalogue. The term “slot” is used to designate the opportunity to extend the ebXML Registry Information Model (RIM).

Azad Faruque asked whether the ebXML specification includes metadata elements addressing the requirements for national security classification, and Joe indicated that it does not include such elements per se but that additional “slots” could be used for that purpose. Owen noted that what are called “slots” in the context of ebXML are called “user-defined metadata” in the context of the DoD Std. 5015.2 for records management applications, and that the first revision of that standard incorporated metadata elements for national security classifications. Owen asked if anyone has built a DoD-certified electronic records management application on the ebXML specification, and Joe responded that while he was unaware of anyone having done so, that is certainly a possibility.

Joe noted that the specification builds on WebDAV and HTTP 1.1, meaning that one ebXML-compliant registry can query another one, thereby also providing the foundation for interoperability with UDDI registries. In addition, the 3.0 specification addresses content validation and cataloguing. Life-cycle management is addressed in the home registry with event notification for updates in other registries. While the specification currently provides only for peer-to-peer communications and contains no notion of hierarchy, Joe suggested that may be a topic of focus in version 4. With respect to the capability for federated queries, Owen asked whether the ebXML Registry TC has had any contact with Eliot Christian, who headed up the search subgroup of the Interagency Committee on Government Information. Joe indicated they had not but that it would be good to make Eliot aware of the new, search-related features of version 3.0. Roy Morgan volunteered to gather and convey to Joe any enhancement suggestions participants in the xmlCoP may have for potential inclusion in ebXML version 4.

Joe said the specification supports subscription to an event notification service. For security purposes, ebXML 3.0 supports XACML and SAML as a service provider, not an ID provider. Finally, in response to a question about support for other metadata besides the elements addressed in the RIM, Joe said that XSLT stylesheets can be used to map other metadata to the RIM.

Joe’s presentation is available at http://xml.gov/presentations/ebxml/registryv3.htm

KC Morris briefed the group on NIST’s interoperability testbed and particularly the Model Development Life Cycle (MDLC) they have outlined. The purpose of the MDLC is to provide a well-specified, repeatable process for developing content standards. Consistent with the MDLC, NIST is developing a suite of model validation tools, including schema qualification and schema compatibility analysis tools, to match schema to data and vice versa. Holly observed that the parser validation tool would have been very helpful when FSA was developing its schema. KC noted that any number of parser validation tools can be added to the service.

KC’s presentation is available at http://xml.gov/presentations/nist4/mdlc.htm

Assisted by Peter Walsh of Plumtree software, Rex Brooks briefed the group via teleconference concerning the results of his derived XML registry prototype, using Plumtree’s Web crawling and indexing software. First of all, he noted the need for searching and crawling standards and cited Eliot Christian’s work for the ICGI. His intention to conduct a live demonstration of the results of the prototype was thwarted by a failed computer, but he did display screen shots of the results. The primary finding was that XML schema indexed on FirstGov cannot be automatically crawled and indexed in an XML registry. Several of the relatively few schema that could be accessed have been posted by the Office of Information and Regulatory Affairs (OIRA) at the Office of Management and Budget (OMB). Owen noted that OIRA administers the Paperwork Reduction Act (PRA) and that the information burden reduction process conducted by OIRA is closely related to the eForms service of the eGov Business Gateway. Peter lamented the fact that the .gov repositories available for crawling did not contain any useful XML schema content.

Rex noted that he had recently learned the interagency Core Data Types Focus Group (CDT-FG) is being disbanded, with its activities being subsumed by two other activities, the DRM working group and the National Information Exchange Model (NIEM). He also noted that the CDT-FG had been working on a glossary of data-related terms. Owen suggested these developments are a positive occurrence since the DRM working group is a governmentwide initiative to flesh out the Data Reference Model of the Federal Enterprise Architecture (FEA) and the NIEM program management office will have more substantial resources than the CDT-FG. Peter reiterated that it was disappointing that it proved impossible to crawl and access information via FirstGov since agency Web pages do not contain good, reliable metatags.

Rex concluded by suggesting that both UDDI and ebXML registries are needed because they address different requirements, and he reiterated that search standards are also required. His presentation is available at http://xml.gov/presentations/starbourne2/derivedregistry.htm

Bob Greeves observed that mixed signals have been given regarding the cost of XML registry services. The business case funded by GSA identified a projected cost of $2.1 million; FSA developed its registry for $290,000 and Adobe suggests that its registry could be implemented for a small fraction of that cost. He suggested those conflicting figures need to be reconciled so that agencies can work together to meet their common needs for XML registry services.

The next meeting of the xmlCoP is scheduled for April 20.

Among those in physical attendance were:

Owen Ambur, Co-Chair

Scott Fairholm, NCSC

Azad Faruque, Department of State

Simon Frechette, NIST

Puja Goyal, NIST

Robert Greeves, DOJ

Steve Hamby, Software AG

John Hanle, proSIMP

Amin Hassam, i411

Biu Henry, CapWIN

Michael Isman, Booz Allen

Larry Jacobs, IRS

Kim Koran, CSC

Serm Kulvartunyou, NIST

Meng Lin, DOT

Tom Merkle, NIJ

Roy Morgan, NIST

KC Morris, NIST

Quyen Nguyen, NARA

Marion Reinson, Starbourne

Russ Ruggiero, HumanML

Sol Safran, IRS

William Taylor, Starbourne

Christopher Traver, DOJ/OJP

John Triplett, IRS

Peter Walsh, Plumtree

Rita Yadav, proSIMP

Those who identified themselves as participating via teleconference were:

Rex Brooks, Starbourne Communications

Steve Rainey (sp?), Transcom

Please convey any additions or corrections to Owen_Ambur@ios.doi.gov