Federal CIO Council XML Working Group
Meeting Minutes, August 15, 2001
American Institute of Architects, Board Room

Please send all comments on/corrections to these minutes to Laura Green.

Working group co-chair Owen Ambur convened the meeting at 9:00 a.m. at the American Institute of Architects. Attendees introduced themselves.

Mr. Ambur informed the WG that he hoped the day's meeting would provide the WG with a clear understanding of the standards that should be included in an XML.gov registry/repository.

Ms. Carnahan began the session by stating that today's speakers would focus on registries. There would be no briefings on individual initiatives, for the goal of the meeting was to begin identifying the standards that the WG should be looking at and should have some impact on. Now is the time that these standards are under development, so the WG can still make an impact upon their development if it wants to. Given that there are multiple standards, part of this session is to understand the complimentary or conflicting roles they may have. We need registries and want them to be standards-based. All of the solutions presented will offer a look at specifications and insight on their interoperability.

Ms. Carnahan expressed a desire for the WG to be able to say that the specifications it chooses will be interoperable with any others it may meet down the road.

Ms. Carnahan then introduced each individual speaker.

Bruce Bargmeyer (EPA and Lawrence Berkeley Lab). Mr. Bargmeyer addressed the characterization of the registry landscape. He informed the WG that there are a number of registries that are of interest to all, but some are more interesting than others. The list of registries includes the ebXML registry, ISO 11179 registries, which differ from the ebXML registry in their focus on the lowest (data element) level of semantics, UDDI registries, database repositories/systems catalogues that are involved in enforcing integrity and running queries, and CASE tool repositories. There are ontological registries that store information on the meanings of words and how separate concepts fit together.

Mr. Bargmeyer stated that the point of this list of registries is that since there are so many out there, he believes that developers should neither try to stuff all of them into one big repository nor create one registry that complies with all. Rather, developers ought to try to make all of these repositories interoperable. These registries differ in granularity, on the amount of semantics management, and on how much they care about meanings. Some are concerned with syntax, others with semantics. But all registries address the same data. These registries need to have points of contact so data can be managed seamlessly. Users should be able to seamlessly manage their data. To do so, developers must find points of articulation and cooperation, and they should find these points over the next year or so.

Mr. Bargmeyer stated that he would like to put together a conference bringing together the people responsible for each registries to discuss how to best to make them interoperable.

David Webber (XML Global). Mr. Webber provided the WG with a review and demo of the ebXML registry. Mr. Webber's presentation is available online at the XML.gov web site in PowerPoint and HTML formats.

Mr. Webber showed the WG a slide illustrating the ebXML Technical Architecture. Companies who use ebXML to conduct their business will interact with one another around the registry. The registry is the focal point of the architecture.

Mr. Webber reminded the WG of ebXML's mission (to enable anyone, anywhere to do business with anyone else over the Internet), and functions (trading profile management and discovery, business process collaboration, exchange of XML business semantic knowledge, and assembly of transaction content). There have been registries in the past, but the ebXML registry is different in that it drives the process.

Mr. Webber then showed the WG a slide illustrating the automated trading partner setup. ebXML enables business processes by providing classification structures, registering business domains, detailing process steps, facilitating and choreographing details, and providing linkage to core components.

Ms. Carnahan asked if there would be any government representation at the X12 meeting in the afternoon.

Mr. Ambur replied that he would be attending. [Editor's note: A summary of that meeting is available at http://www.x12.org/summit/.]

Mr. Webber explained the legal role of ebXML: it is a service that manages and secures business interaction information in a consistent way that is compliant with emerging international standards and technical specifications. The legal role of ebXML must be considered, as ebXML will be around for quite some time.

Mr. Webber then showed the WG a series of slides illustrating how ebXML meets the integration needs of businesses. ebXML will build upon existing business infrastructure and build an information services layer in business infrastructure. The government owns a great deal of legacy data, so it would create an XML layer above that data. Not all of the data in the legacy layer is data one will want to interchange with and expose to the outside world. Thus, an open socket interface is needed. This is where ebXML, UDDI, and other solutions come into play, with the goal of providing a single face solution, single interface through the information services layer. There must be integration at the point of use. Developers need to derive core components and standard business processes to reduce interface back-end complexity.

ebXML's method for reducing the number of wires at the back end is bizcodes. In the ebXML specifications, these bizcodes are written as UID. Bizcodes link physical attributes to properties of entities. If one thinks of these entities as items in a store, a bizcode is analogous to an item's barcode.

Kevin Campbell asked what ebXML does with the fact that the meaning of something such as "part number" can vary from organization to organization.

Mr. Webber replied that the registry helps with this confusion, to drive people to common definitions.

Mr. Campbell asked how ebXML can convince large companies to define common touch points.

Mr. Webber replied that while terminology may not be the same from organization to organization, the UID will be.

Mr. Campbell asked if these UIDs could be extracted from trading partner agreements.

Mr. Webber replied that they could.

Mr. Webber then showed the WG a slide describing the relationship of and use of UIDs and a slide showing a model of a core component.

Mr. Webber announced that he would be showing the WG a cut-down version of the NIST interoperability prototype registry. A slide illustrating the registry's concept architecture shows that it contains both ebXML and SOAP-based syntaxes. This registry is online at http://nist.xmlglobal.com/.

Mr. Webber then commenced with a demonstration of the registry. This demonstration is part of the eMall system completed last year. A user can enter the site select the registry and syntax he wishes to use and the UID he wishes to query.

Theresa Yee (LMI) asked how the user would know which registry to designate.

Mr. Webber replied that at some point, the user would have to have a configuration option that would tell him what registry he has hit.

Alesia Jones-Harewood (DISA) asked how the user would know which UID he was looking for.

Mr. Webber responded that that is the purpose of the registry. At some point, a business analyst has gone into the registry to create a DTD for a transaction. He can look up a UID using the registry's classification system.

Ms. Carnahan added that the projection of the demo in the meeting is not what a typical registry interface would look like. Rather, the view provided by the demo is more focuses on showing WG members the nuts and bolts of the registry. Actual registry users would only see the inquiry screen.

Mr. Webber remarked that the user can query associations in the registry.

Mark Crawford (LMI) noted that in traditional business transaction exchanges, an organization must map to whatever its trading partner wants to send it. This mapping must occur data element by data element. With ebXML, mapping has been moved from the proprietary back-end mode to a public mode. Under this concept, organizations still have to do proprietary mapping, but because they are moving it into the public mode, they only have to do it once. They can then transform it into any standard or syntax the trading partner wants to use. The UID and core components are syntax neutral. They can be used for any syntax, not just XML.

Joe Chiusano (LMI) asked if it would be possible to free oneself from registries.

Mr. Webber replied that developers must address the issue of design time versus run time, and at run time, do they want to rebuild the registry every time it is queried.

Eliot Christian (USGS) commented that it appeared as if Mr. Webber was saying that by attributing, the user is being sent off to one syntax where only one thing can apply.

Mr. Webber responded that that there are extensive definitions of core components. One thing learned from the NIST prototype was that people are fixated on minimum and maximum lengths. These are not important; what is important is the business semantics side. The UID mechanism allows one to slice at the core component and get the pieces of it one wants. This sets ebXML apart from X12.

Lex Poot (DTS) remarked that this relates to the RosettaNet initiative.

Mr. Webber replied that RosettaNet will be using an ebXML registry.

Chuck Hazel (DISA) inquired as to who owns/controls the UIDs.

Mr. Webber replied that the ebXML specifications allow for organizations to create their own UIDs right now. Ultimately, however, there needs to be a decision as to how to distribute ownership.

Mr. Crawford added that the core set of core components are the ones that are being worked on by a UNCEFACT group.

Mr. Webber then discussed next steps for the NIST prototype. There must be a bridge to the legacy systems. This process is illustrated in one of the presentation slides.

Mr. Webber then discussed the differing missions of four different kinds of registries: ISO11179, ebXML, OASIS, and UDDI.

The ISO11179 registries are rooted in the EDI world, and are centered on data dictionary synchronization and metadata registry harmonization across large enterprises. OASIS registries find their heritage in the SGML world. Their focus is on mixed content information models. ebXML registries are focused on A2A exchange. UDDI registries provide business directory services and automated trading partner discovery options.

Mr. Webber then showed the WG a slide displaying a brief gap analysis.

Mr. Webber summarized his remarks by stating that there is a need for a semantic registry, a human interface, and a machine interface. There are still several issues which must be addressed, including classification and associations, UID and content notation synchronization, registry of registries (UDDI is an option for this), and certification of registry implementations.

Mr. Campbell remarked that each industry has a set of proprietary semantics. In the EDI realm, an organization needs to find the semantics for each representation based on person, device, and industry. There are also significant cost issues in the EDI realm, where building registries can cost billions of dollars. ebXML can create a bridge between what EDI really is and what people think EDI is.

Mr. Webber pointed out that the government can play a vital role in dealing with UIDs. The government has a stake in making sure things are done correctly.

Anne Thomas-Manes (UDDI.org). Anne Thomas-Manes then gave the group some comments from a UDDI standpoint.

Ms. Carnahan asked Ms. Thomas-Manes to discuss the relationship of UDDI to other registries. She stated that she perceives UDDI as addressing directory services while ebXML focuses on business processes. The coupling between the two occurs when a user finds a service using UDDI, then drills down into the UIDs with ebXML to identify possible crosswalks.

Ms. Thomas-Manes remarked that she distinguishes UDDI and ebXML by the fact that UDDI is simply a registry, not a repository. UDDI is an index of things that will point the user to a location.

Mr. Christian asked if this meant that the user would have to deliberately switch information models from UDDI to ebXML halfway through the information seeking process.

Mr. Webber replied that the user would not have to switch models, but would simply move down through different layers of information.

Mr. Christian asked if the process could work the other way, if an ebXML layer could be laid atop a UDDI layer.

Mr. Webber replied that there is currently work going on to accomplish this.

John Dodd (CSC) asked if an organization were at the front end of a collaboration process, would it use UDDI as a first level to discover who the various players are. He wondered what the process would be.

Mr. Webber replied that an organization could UDDI in that manner. He cited an example from the Canadian government. In order to establish a restaurant in Canada, the owner must fill out twenty different forms. It is only logical to have a single point to go to find the departments the owner must register with and then channel into those despartments' interfaces. The WG needs to ask itself how best to take government as it is today and expose it electronically.

Mr. Dodd commented that people usually do not care where the information they need comes from, they just want to get it.

Ms. Thomas-Manes added that not every service needs ebXML to support it. There are many lightweight services that will be supported simply by UDDI.

Mr. Webber states that the key question an organization must ask itself is if its service is going to lead to a long run thing. Recurring things drift over into the ebXML model.

Chris Kurt (UDDI.org). Mr. Kurt began his remarks by asking what the WG wanted to leave the day's meeting with.

Mr. Ambur replied that the WG should be looking for a consensus on what the Federal government should be doing with its registry, identifying the low hanging fruit, and establishing its priorities. The administration has formed a citizen's center eGov task force to come up with recommendations to serve as the foundation for money to be spent on eGov funds. In some ways, a registry/repository is of the essence. He hoped to come out of the day's meeting with a consensus as to what that should be.

Mr. Kurt remarked that we should be clear that what has been presented so far in the day's meeting is one version of what potentially could be, not a reflection of how things are today. He felt that the preceding discussion had been presented as a statement of exactly how registries work. However, Mr. Kurt argued, not all of those are characterized completely accurately. It would behoove the WG to think more broadly than EDI and B2B. All ebXML documents refer to ebXML as an EDI solution.

Mr. Crawford interjected that the ebXML requirements statement never said that ebXML is an EDI solution.

Mr. Kurt stated that the WG needs to go beyond thinking in terms of B2B to think of what kind of electronic services are provided ad infinitum to other agencies and citizens. There is a broad set of behaviors that can be described by UDDI.

Mr. Webber added that industries say that B2B is a drop in the bucket; they are more concerned with A2A solutions. Businesses are going to go after facilitating A2A systems as ROI. They believe that there must be an ROI in order to change. If there's no ROI on a change, then there's no need to change.

Mr. Kurt stated that if the goal of the day's meeting is to leave the room with a recommendation as to how these part should fit together precisely, that goal will not be met.

Mr. Webber comment that the WG should focus on coming up with ideas as to what some of the key pieces of the registry/repository should be and where those pieces might fit together.

Mr. Kurt offered one final suggestion, that the WG divide its conversation into three areas of functionality: initial discovery mechanisms, repository storage of artifact mechanisms, and the semantic business content mechanisms.

Mr. Webber proposed a fourth area: the API and socket ware.

The WG then broke for ten minutes.

The second half of the day's meeting was dedicated to a panel discussion of registry standards. The panelists were Mr. Bargmeyer, Ms. Carnahan, Mr. Hazel, Mr. Kurt, Ms. Thomas-Manes, and Mr. Webber.

Mr. Bargmeyer began the discussion with a brief update on ISO 11179. ISO 11179 is under revision. Part Three, which contains a metamodel of things one would like to know about data, is in its final committee draft and should be finalized by November. There are substantial efforts underway in implementing 11179 registries. There is an implementers' coalition and there are registries in the aviation, health care, and environment industries. ISO 11179 registries have an EDI and database flavor. They are useful in managing semantics sot that people entering data can achieve some kind of consistency.

Mr. Webber added that ISO 11179 registries are focused on an entire enterprise, not just one piece. These registries will go to every last closet and nook and corner to find data to enter into the registry. One can contrast this with ebXML, which deals only with the information that an organization wants the outside world to see.

Mr. Bargmeyer pointed out that ISO 11179 registries have a strong flavor of coming to a consensus about data. A good deal of the cost in implementing such a registry comes from trying to bring users to a consensus.

Mr. Webber asked if Part Three of ISO 11179 would have a machine API.

Mr. Bargmeyer replied that a machine API is not in Part Three, but the ISO folks are working on one.

Mr. Hazel then spoke about DISA's registry. Four years ago, DISA started to notice that XML was being applied to Department of Defense (DoD) systems. However, these efforts were disconnected, with each team developing its own elements. DISA realized that these disparate implementations would cause a good number of long-term problems. The DISA registry was established to provide a marketplace for harvesting as much of that data as possible and providing a place where it could be exposed to other developers. The focus of the registry is on harvesting what data concepts developer actually need and putting them out where they can be seen and used. Currently, there are over 15,000 concepts registered and 12 namespaces residing within the registry. The registry has a strong bottom-up focus, and is focused more on the ISO 11179 standard. DISA is looking at developing more of the etymology needed to begin mapping between constructs. Currently, the registry has only a user interface, but there are plans to put a machine interface into the next version, which should be out in FY02. The DISA registry is not an eBusiness based registry. It is integrated and working with commercial XML groups and consortiums.

Mr. Ambur remarked that he liked the DISA model, because it seems to be the only one that works. If one thinks of the Federal government as a whole, it is impossible for anyone to impose their will on the entire bureaucracy. But once the WG registry has been demonstrated to be usable and scalable, it would be reasonable for OMB to direct agencies to use it to advise others what they are doing with XML and, in turn, for others to use it to subscribe to the data elements that are applicable to their functions.

Mike Todd (OSD) noted that the GAO is currently reviewing XML implementations across the Federal government to discover the best way to support interoperability in order to help guide policy and potential legislation. Members of the WG should be aware of this effort and be on the lookout for the final report, which should be available in February of 2002.

Bill Morgan (GSA) commented that it seems that within the government there should be a framework that is not driven by registries or repositories. There needs to be more communication between sectors as to what they are doing and why and how each effort fits into a larger framework. There needs to be a plan outlining the government's short- and long-term priorities. Otherwise, the government will spend a lot of resources trying to undo things that did not need to be done.

Marion Royal (GSA) responded that perhaps the XML.gov web site could act as a central clearinghouse for white papers, etc…. There is a need for a resource that is obvious to the casual and not-so-casual browser.

Mr. Hazel outlined the mechanism DISA uses for managing its registry. Namespaces represent communities of interest within the registries. The managers of these namespaces are not necessarily DISA employees. There is a forum where the namespace managers can get together to decide how the registries should be managed. Experience has shown the problems with a top-down approach. Developers who are under pressure to develop a product will use whatever technology is appropriate regardless of what they are told to use from the policy level. We need to make it possible for developers to coordinate with everyone else and work together.

Ms. Jones-Harewood added that DISA was asked by the DoD CIO executive board to prepare an implementation plan for the registry. DISA had intended for its registry to have a narrow focus, but the CIO board asked it to extend the focus. DISA would appreciate any feedback from WG members.

Matthew Kern (Productization) commented that Mr. Morgan's point was cogent. Implementation of this technology is immediate. The government has a role in standardizing but no particular role in competing with industry. The government is under no mandate to implement. However, we are on the verge of forcing out all small business from this technology. Any recommendation to the eGov task force should include funding to help small businesses adopt this technology. The government needs to focus long-term resources on long-term problems. As for standardization, we need to collapse many efforts into a unified effort that focuses on standardization, not on implementation.

Mr. Webber responded that in the standards process and the work being done in ebXML, the challenge is to create a product that bridges technologies and makes it possible for businesses to use it. Very lightweight coupling mechanisms like the UID can help. The marketplace will drive the development of specifications. The methodology is much more important than the immediate technology.

Mr. Kern agreed that the marketplace feedback mechanism will work, but we need to make sure that the feedback methods are supporting small businesses.

Mr. Hazel noted that the National Imagery Mapping Agency's Open GIS Consortium has defined the vocabulary for that domain. We find that the big players in defining those standards are not the big companies, but the small companies who are trying to carve out a niche in an industry dominated by two or three big players. By getting their requirements written into the specifications, the Consortium is supporting small business.

Mr. Christian commented that UDDI, ebXML, etc…are all of the same nature. They are not standards, they are emerging technologies and the government cannot come to a consensus on them. It needs to build a testbed to try them out. The immediate goal should be to build a testbed to show the power of these technologies.

Ms. Carnahan suggested that rather than coming to a selection , there are certain efforts within each of the four areas previously identified that should be mapped out. These efforts should be identified as emerging efforts to track. In terms of small business, the reason why there are standards is to protect them. NIST tries to push standardized solutions so that if the interfaces are there, everyone has an opportunity to compete.

Mr. Bargmeyer reminded the WG that the four areas are discovery, repository, semantics, and socket layer.

Rick Smith (MPG) told the WG that the day before the meeting he spent some time with two major contractors discussing a virtual data broker and other names for registries and how to set up a procurement system with the 16 legacy systems the registry would feed. The top data gurus in the world are not sitting with the WG. The gurus could care less about XML, to them it is just a blip on the map.

Ms. Thomas-Manes commented that it sounds as if DISA's work has been solely registry- and not repository-focused.

Mr. Hazel responded that the DISA registry is fundamentally a repository of concepts with no registry behind it.

Ms. Thomas-Manes inquired as to DISA's classification scheme.

Mr. Hazel replied that the degree of classification is dependent upon the classification inherent in the namespace. The namespace experts are the ones who decide if they are going to break things down. In the next version of the registry, the elements do not exist as unconnected unrelated elements. If a user can go into the registry and identify elements, then he can trace that taxonomy through the rest of the registry. The definition of what that taxonomy is supposed to be is left to the namespace manager.

Ms. Thomas-Manes then updated the group on UDDI. UDDI is going along very strongly. Version Two has been released, and the requirements list for Version Three has been finalized. There are a number of implementations of UDDI available: IBM has a server, and there are many other open source products. This technology is moving along very quickly. There are not yet a large number of services registered in UDDI. There is not a lot of acceptance yet, but the initiative needs to get structured tools in place before it can see massive adoption. Ms. Thomas-Manes worked with IBM to put together a paper for ebXML using UDDI as a registry of registries, and she is looking to revamp this paper. There is a need to devise a set of UDDI models for the ebXML registry. A groups within Sun Microsystems is developing a UDDI "best practices" paper.

Mr. Christian added on to the government context of UDDI by informing the WG that there are 1500 records for agencies in the Federal government. Professional classifiers at the University of Chicago are providing classifications. They have registered services for agencies.

Ms. Carnahan asked if record of any of this work is currently available on XML.gov.

Mr. Ambur replied that it is not.

Mr. Christian informed the WG that there will be an update on this work on August 24.

Ms. Thomas-Manes added that it would be nice if anyone working on schemas or related initiatives in industry would register those within UDDI.

Mr. Christian added that the models for geo-spatial and GILS access have been registered.

Mr. Webber asked how the registration in UDDI works.

Ms. Thomas-Manes replied that there are two separate kinds of items that go into the UDDI registry: a service description and a service type.

Mr. Webber commented that that works nicely because a PIP is a process definition. So when one speaks about a schema, one is speaking about what the PIP looks like. This is different from BizTalk.org, where one registers schemas as schemas.

Mr. Kurt added that the notion of a t model is a behavior or any sort of item that can be specified or implemented consistently. With UDDI, the user can register service behaviors, but usually there is some metadata that goes around the behavior that must be formalized.

Mr. Webber remarked that the schema itself is the physical mechanics and a t model is metadata about metadata.

Mr. Kurt said that there is a unique identifier that allows users to have some composite description of what a service is.

Ms. Thomas-Manes stated that within the Java community, there is work ongoing to develop a paper on Java APIs for XML registries. There is a desire to find a single data model to represent data in a registry. This will be done through the Java community process.

Mr. Webber remarked that within OASIS and ebXML there is work ongoing on information models for exactly the same thing. The two efforts should look for points of connection.

Mr. Kurt told the WG that UDDI published its Version Two specifications a few months ago, and the implementation should be available in September. An interesting feature in Version Two is the ability to extend identification taxonomy schemes to include an arbitrary set describe by a third party (e.g., government).

Mr. Ambur asked if the FirstGov folks are aware of UDDI.

Margie they are looking at that as a prototype of a Federal portal. They are now figuring out how to do a fully thought-out portal to the federal government. They would like to bring together collaborative XML work done in the UK.

Mr. Kurt added that this forum tends to focus on XML, but those services that are non-XML are important, as is laying the groundwork for this type of technology by starting to register and surface non-XML behaviors.

Mr. Kurt then provided a brief update of BizTalk's work. The BizTalk framework, a set of guidelines for implementing an XML schema and a set of XML tags used in messages sent between applications, has been subsumed into much of the W3C's work.

Ms. Carnahan then updated the WG on ebXML. The ebXML TC Version One in May formed a number of subteams. The whole issue of cooperating registries and how they can talk to each other and make use of the information shared is of concern. There are several meetings upcoming in September and December. Any documents ebXML authors are available online.

Ms. Carnahan told the WG that the big thing that came out this meeting is that the WG is not coming up with a list of specifications and is not yet ready to do so. She likes the notion of breaking work down into the four areas proposed and trying to define the specifications within those areas. She would like to have a group of people take a list of specifications and bring back their recommendations to the group via the mailing list.

Mr. Ambur suggested that the WG address both short-term and long-term goals.

Ms. Carnahan remarked that there is a bit of a chicken and egg problem. This is a cyclic process in which something needs to be built as a starting point that the WG can learn from. She stated that a COTS solution would be the best avenue to pursue.

In terms of the bigger picture, Mr. Ambur noted that Mr. Royal has been detailed to the citizen-centered eGov action team, whose recommendations will provide the basis for the Administration's priorities for the expenditure of eGov funding.

Mr. Kern stated that some of those funds need to be directed to small business to keep up with the standardization effort.

Ms. Carnahan said that it would be helpful if, in outlining the WG's effort, the WG can also outline what the standards do. That way, the WG can say, "here is what this gives you and here is what you can do with it." Then the people who need solutions will be better able to find them.

Mr. Kern commented that ROI and requirements should drive the entire initiative. However, focusing the money on further entrenching an architecture that competes with industry is not the mandate of the WG. The mandate of the WG is to amp the requirements so business can get in sync with standards and not with architecture.

Mr. Kurt recommended some steps for the WG. Looking at the four general areas, they appear to be on different schedules and have different sets of application technology. For discovery, UDDI is the place to start to give everybody an opportunity to catalog whatever XML/non-XML objects they have. This can be done in the immediate term. For repository, there are a few options that are not quite along in the technical side, such as ebXML (for XML artifacts) and others that exist internally (for non-XML artifacts). For semantics, ebXML and other industry standards will work. This is a much longer-term activity that will require more resources. For API-type interfaces, he suggested that the organization defining the build of this work be represented here: the W3C.

Mr. Hazel added that this meeting has received a description of two different types of registries: service registries and data concept registries. The WG needs to distinguish between the two and treat them differently.

Mr. Webber said that he could see ebXML serve to work with business process artifacts. From a government point of view, methodology and framework are very important. The government needs to adopt a position where things are not business as usual.

Mr. Kern commented that the people who are the fundamental drivers of workflow in this process are not represented on the WG. They should be.

Mr. Ambur informed the group that next month's meeting will discuss how XML can be used to enhance DoD 5015.2.

Mr. Webber commented that there is a good deal of work to be done in that area.

Mr. Kern asked where the funding comes from for small businesses to attend forums such as WG meetings.

Mr. Ambur replied that ANSI has scheduled a forum to consider how XML can be used to improve the process by which standards of all kinds are developed. [Editor's note: Information on the forum is available at http://www.ansi.org/rooms/room_5/public/XML_Forum.htm.]

Ms. Carnahan said that she would be willing to take a shot at trying to write the first concept for the four areas. She would appreciate any volunteers to assist her.

Mr. Poot pointed out that the WG needs to address legislative issues.

Mr. Ambur concluded the discussion by saying the WG will be looking to NIST for guidance and leadership on standards-related issues, while GSA will be handling operational issues. Finally, he asked each of the speakers to review the minutes carefully to ensure the meaning of their remarks has been accurately recorded.

Mr. Ambur made the following announcements:

The XML SGML User's Group will meet August 15 to discuss XML intelligent content firewalls.

The xml.house.gov web site has gone public. Please check it out.

The Logistics Management Institute's tag study has been posted on the XML.gov web site. WG participants are invited to review it and comment on it on the listserv.

Next month's WG meeting will focus on how XML can be used to enhance the DoD 5015.2 requirement for records management. The October meeting will focus on a discussion of XML Query. Pat Case from the Library of Congress will facilitate the agenda for the October meeting.

The WG's strategic plan is due by the end of September. If anyone has suggestions for the plan, please inform the co-chairs or post them on the listserv, bearing in mind the fact that the WG is not a policy-making group and has no ongoing operational authority.

Mr. Crawford announced that the EPA has commissioned a study on registries and has received a report that addresses many of the issues that agencies should consider when implementing XML. If anyone is interested in learning more about that study, please contact Bruce Bargmeyer or Steve Vineski.

Due to the disappearance of the sign-in sheet, there is no record of attendees for the August meeting. Additionally, the absence of a written record of names may have resulted in misspelling of participants' names in the minutes. If you catch any mistakes, please let me know at lgreen@lmi.org. Thank you!