Federal CIO Council XML Working Group
Meeting Minutes, June 20, 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.

Paul Daisey (Census Bureau) and Kurt Buehler (Open GIS Consortium) briefed the WG on Geographic Markup Language (GML). This presentation is available at the XML.gov website in PowerPoint and HTML formats.

Mr. Buehler began by providing the WG with some background on the project. OGC stands for the Open GIS Consortium, a non-profit trade association founded in 1994 to promote interoperability amongst systems that process geo-referenced data as well as between these systems and mainstream computing systems. OGC is an international organization with 204 member organizations.

There are many different paths to standards. Mr. Buehler showed the WG a slide that depicts these various paths. The left side of the slide depicts a very common path. The end goal of each path is always an international standard. That is why OGC is an international organization. In fact, many countries will not codify a standard unless it is already an international standard. OGC works with the .com, .edu, and .gov communities and has a collaborative relationship with the International Organization for Standardization (ISO).

Mr. Buehler then showed the WG a slide outlining OGC's specifications and traditional development sequence. The development sequence begins with an abstract model usually expressed in UML and English. OGC moves through this model to end up with interface applications. OGC is about interfaces. The specifications on the slide are engineering-level specifications. The features are simple; there are gridded coverages and distributed catalog services. Additionally, OGC has formal coordination with ISO TC211 that allows it to work in joint WGs with ISO TC211 and submit its specifications as draft international standards. OGC's goal is to promote the development of standards based COTS software. OGC works with its vendor and user members to try and promote the development of this COTS software.

Mr. Royal asked if Mr. Buehler could describe his definition of "catalogue" in a little more detail.

Mr. Buehler replied that he was speaking of an interface. When OGC began its modeling work, it used "catalogue" to refer to a repository of metadata entities. Now OGC's use of "catalogue" is more of a legacy name.

Mr. Royal asked if OGC is using any industry standards for its repositories.

Mr. Buehler replied that since OGC's specification was adopted several years ago, it uses whatever standards were available then.

Mr. Buehler then presented the WG with a diagram of Enterprise GIS in an OpenGIS Environment. This works on a framework of open-faced GIS interfaces that allow for plug and play with COTS. OGC's view is that one can model one's work from a technical and a market point of view. The goal is to be able to utilize these interfaces and standards-based commercial solutions to build systems that one can grow and live with.

The OGC specifications for the web followed a traditional standards development process through 1995. However, the speed of development was rather slow. Accordingly, OGC wanted to figure out a better way to coordinate its user community with its vendor community. To do this, OGC invented a "web mapping test-bed" that brought government sponsorship together with vendors in a hands-on mapping process. OGC wrote specifications that it then prototyped, integrated, and demonstrated at much faster pace. Out of this effort, OGC broadened into a program that now exists as the Interoperability Program wherein the consortium puts together sponsors with tech providers in a collaborative process. This year, OGC has kicked off the OGC Web Service Initiative where the consortium will look at new and emerging web interfaces.

There are several problems with finding maps on the web today. There are many Mapquest-like sites out there, all of which have some problems. The user wants to know if there is any simple method by which he can request a map from a server and display it in his browser. Currently, there is no interoperability between the web and the data servers. OGC has a technical vision for this in which a user can discover, access, and retrieve views from multiple servers all over the web and display multiple layers of geographic data within a single web application. To do this, one has to make sure that the data lines up (e.g. is in the same coordinate systems) and that all other kinds of data are available. Mr. Buehler then showed the WG a screenshot illustrating this kind of web mapping ability.

The results of Phase One of the Web Mapping initiative are a four map server with HTML and XML services Special features such as getCapabilities, getMap, and getFeatureInfo allow the user to learn the capabilities of a server, retrieve a map from it, and retrieve information on a highlight feature of a map.

Janina Sajka (AFB) asked for an example of pointing.

Mr. Buehler responded with the example that if a user made a request of an image with roads and buildings overlaid, the getFeatureInfo request will tell him that he is pointing at the buildings layer and provide information as to what those particular buildings are.

OGC has run multiple test-beds, all of which are collaborative. Part of its development process involves the development of what OGC calls "Candidate Specs." These are specification that the user community is allowed to test drive. These specifications are then brought back to OGC with comments before release as specifications.

Mr. Daisey then showed the WG an illustration of the OGC's Web Feature Server (WFS). The WFS grew out of a map server and deals with data rather than maps. The WFS will list for the user the different feature types it can serve him. This information is returned in the form of GML schema. This information allows the user to specify the features he wants to use/receive. The user then receives a collection of the features he specified in the form of a GML document. The user can lock a feature, and this ability allows him to conduct a transaction. The data exchanged here underlies the vector-based geographic information.

Mr. Daisey then showed the WG a slide listing GML's initial design goals.

The GML project was initiated by the OpenGIS Web Mapping SIG. The evolution of the project is displayed on the "GML Evolution Slide" at the XML.gov website.

Mr. Daisey then showed the WG a slide describing the structure of a GML document. Each GML document contains a set of Feature Collections. A Feature Collection is a GML Feature. Complex features are composed of more than one feature. Each Feature is composed of properties, one or more of which may be a Geometry. A GML Geometry contains coordinates and specifies a spatial reference system.

One thing OGC came away with from Version 1 of GML was a document that serves as a content model. The structure of this model is shown on the "GML Content Model" slide. The idea is to have some forward compatibility with the future semantic web.

Mr. Daisey then showed the WG a diagram of the GML v2.0 Geometry Schema. The right half of the slide describes a Geometry Collection.

Mr. Daisey then showed the WG a similar diagram of GML v2.0 Feature Schema.

Mr. Daisey then outlined the W3C's schema recommendations. He stated his belief that schema will replace DTDs in the data case of XML, but maybe not in the document case.

Mr. Daisey walked the WG through several slides outlining the differences between XML and GML using an example of a document describing a university dean. This is illustrated in the "Encoding GML Features Without Geometry" slide. Red text indicates inherited attributes from the GML feature type. In the following slide, one can see how the <point> tag can be used to define a coordinate. A simple geometric property is defined in an element. This property could be a point. In the university dean example, the dean has a location in space and GML properties help one keep track of where he is.

Mr. Daisey then switched to another example using a city model to illustrate the use of two primary elements: rivers and roads. A city model is a collection of elements like these. These help define GML Feature Collections.

The "Encoding GML Feature Collections" slide shows what the actual data will look like. Remote Association Members are another kind of data not displayed in the document on this particular slide.

The following slide, "Distributed GML Feature Collections," shows an example of how to combine data from multiple servers. Data such as name, date, storm, county, etc…are all examples of the kind of data that can be collected from separate servers. This particular use case has not yet been pushed out. It does not represent an actual project yet.

OGC is currently working on GML Version 3.0. OGC has adopted ISO 19107 as the geometry and topology model. OGC will incorporate ISO 19115 metadata (a standard for geographic metadata). Version 3.0 will be backward compatible with Version 2.0. OGC is also going to try to incorporate specifications from concurrent Japanese initiatives.

This concurrent Japanese initiative is called G-XML. G-XML is a coordination project initiated by MITI* that includes participation from government, academia, and the private sector. All involved have agreed to base as much of G-XML on GML.

GML Version 3.0 has a Revision Working Group. This group is working on incorporating two features from G-XML into GML: Spatial Locator and Location of Interest.

In April, OGC laid out a "To Do" list for GML v3.0. OGC hopes to publish GML v3.0 in December. Please note on the "GML 3.0 Schedule" slide, "September TC" should read "October TC."

Mr. Daisey then briefed the WG on a potential use for GML: using GML schema on U.S. Census TIGER/Line files. TIGER/Line is an existing file format that stores some Census data that begins with line pieces. The series of slides beginning with "Prototype GML v2 Schema for U.S. Census TIGER/Line Files" illustrate the relationship between kinds of data.

The translator between data formats is the Open GIS Interoperability Program 2000 Task. More information is available online.

OGC is working on an interoperability project. One case in this project focuses on having a browser-based client that allows someone to do updates to his geographic data over the web. In another case, OGC is looking at pushing quite a few of OGC vendors to develop GML capable interfaces in their GIS products.

Ms. Sajka asked if OGC has any interest in building models for multi-tiered structures.

Mr. Daisey replied that the only place that he knows of such work occurring is in Holland. Those developers have shown some interest in GML and OGC and they are talking about having three-dimensional coordinates.

Mr. Buehler added that GML v3.0 will handle three-dimensional coordinates. Whether or not this capability makes it into commercial software is another story.

Bruce Hoglund (DLA) asked if the USGS is using GML.

Mr. Buehler replied that OGC is running a pilot with WebMapServer and is going to evolve to the web server approach with EROS in the next year.

The WG then broke for ten minutes.

Betty Harvey of Electronic Commerce Connection, Inc and the Washington DC SGML/XML Users Group briefed the WG on accessibility standards in recognition of the week's deadline for Section 508 compliance. This presentation is available in HTML and XML formats.

Ms. Harvey stated that her presentation is intended to focus on the information. An organization's information must be robust enough to be accessible to everyone. The presentation itself was displayed to the WG in XML.

Every person has some sort of disability, be it a physical one, a learning one, an equipmental one, or a language one.

The most important thing an organization will do when developing its XML is analyze its data. This is the most costly part of an XML project if it is not done correctly, for it will have to be done over and over again. An organization must analyze its information for relationships between data, potential users, media requirements, and other considerations. Information should be analyzed for reusability, componentization, output requirements, and granularity.

An advantage of using XML is that it provides multiple outputs from one data source. The user has a DTD or schema that controls the data, and with the style sheet used, a user can output it to a CD-ROM, the Internet, paper, Braille printing, and much more.

Ms. Harvey then discussed the process through which an organization can develop XML to convey its data. The organization must analyze its data while being cognizant of its requirements. For example, if audio presentation will be used, then there must be clues that the sound is there so deaf people will know to look for the same information elsewhere. Then the organization must develop its DTD or schema. To fit the required business model, the organization must define the DTD to support disability requirements. If the organization's data is not robust enough to begin with, it will have serious problems meeting these requirements.

XML can be processed and/or translated automatically to support accessibility issues.

The International Committee on Accessible Document Design has been in existence since the late 1980s. It developed an SGML standard for doing automatic translation from SGML to Braille or voice presentation. It is mainly used for Braille. This is an ISO standard and is widely used. The tags used are very HTML-like. ICADD is necessary because Braille is cumbersome and difficult to produce, and voice is almost impossible to produce from traditional publisher's files. Additionally, the small market for Braille translates into a small supply.

Ms. Harvey then showed the WG a list of ICADD attributes. Again, they look very much like HTML.

Another way to increase accessibility is through the use of voice synthesis. Old speech synthesis systems have the drawbacks of being monotonous and emotionless. They require a long time to get through messages, and the listener cannot easily jump to the "good" part of the message. He cannot filter out the drivel to get to the point easily.

Yuri Rubinsky was one of the first people to begin to address these problems. The W3C just came out with a draft on speech synthesis for use with voice browsers.

Ms. Harvey then showed the WG an illustration of the DTD. At each level in the DTD, the developer can dictate what language he is working in. Ms. Harvey pointed out some of the interesting elements in the DTD. They include "say-as," "phenome," "voice," "emphasis," and "break." All serve to create a more realistic reproduction of a human voice.

There are several other interesting standards. Wireless Markup Language is being used to develop applications over wireless communication networks. Voice XML is another voice recognition standard.

Joe Carmel (U.S. House of Representatives) asked if applications of any of these standards exist on the web.

Ms. Harvey replied that there are some voice browsers in existence, but she is unaware as to whether they are using speech yet.

Ms. Sajka pointed out that the key is the ability to re-purpose content. To do this, a developer needs to know what the data is and what its semantic value is. It would not make sense if everyone had to be an expert in Braille, etc…. An organization must decide who is responsible for this knowledge. This is the key to 508 success.

Ms. Harvey added that an organization needs to have the right repository so it can display the right source at the right time.

David Ravichandar and Cathy Lockwood of Oracle briefed the WG on the use of XML in the 2002 Economic Census.

Mr. Ravichandar informed the WG that he and his Oracle colleagues were here primarily to discuss the eCommerce 2002 census project entitled "EMR" (Electronic Metadata Repository) as well as how Oracle is using XML to dynamically generate questionnaires on the web using metadata.

Mr. Ravichandar gave the WG some background information as to how the EMR project came about. There are several ongoing efforts within the Census Bureau in which people are using XML to try and map to many questionnaires.

A CMR (Corporate Metadata Repository) is important because documentation for systems and data too often exists only the heads of the system developers. There needs to be a way for people to be able to store this information and share it across the Census Bureau.

One of the primary goals of the EMR is to help in sharing and organizing this information. The purpose of this briefing is to illustrate how the team members are generating survey instruments on the web using XML.

The team hopes in the next phase of the project to load geographic maps into the CMR. The EMR has become very popular with Census personnel. The team is actively pushing XML to see how it can be extend to the maximum level to enable reuse and help generate applications on the fly with a web browser. There are tons of metadata documents lying around that can be used for this.

The EMR consists of four components. All of the metadata is stored in the database, which is based on the ISO 11179 standard. The user interface enables analysts who use the program to work with the EMR. The XML interchange is the actual exchange of XML documents as forms are developed. There is also a generalized interment design tool that is a third party tool that exchanges customized DTDs.

The user interface also provides security and depends very heavily on the data in the registry.

Mr. Ravichandar then described EMR XML interchange and showed the WG a slide illustrating the exchange of information. Information comes from several different servers throughout the nation and from a wide variety of operating systems. There is a lot of information going back and forth as XML, and so this interchange is a very complex process due to a lot of on-the-fly validation. Much of this validation is done in the code and at the application level.

XML interchange supports many different content types. Mr. Ravichandar showed the WG a slide listing the sections that the team has broken the questionnaires into. On the afternoon of June 20, the team will deploy the standard question section in an economic census form.

The generalized instrument design tool was designed and developed by Fenestra Inc. and is used as an auto-formatter and forms designer.

Mr. Ravichandar then showed the WG a screenshot of a sample questionnaire with an element entitled "standard question. "He then showed the WG a slide displaying the XML behind this screenshot. The entire XML document is made up of two sections. The first is the control section, which contains transaction ids, parameters, and request status. The second is the data, the result of the database query, which came back from the EMR.

Mr. Royal asked where the naming convention for the element names originated.

Mr. Ravichandar replied that the XML document is based on the database elements where the metadata stored. Since the database in this instance is an Oracle database, the names comply with Oracle specifications.

Mr. Ravichandar then gave an EMR preview demonstration. The screenshots are available in the presentation posted on the XML.gov website.

Mr. Daisey remarked that it looks as if the EMR was set up to do paper forms design. He wondered if the team had thought of developing interfaces for turning those forms into web forms.

Mr. Ravichandar replied that the team has just begun discussing web forms. They are looking into web forms, as doing so is the logical next step for the project.

Mr. Ambur asked if the project is using the XForm standard.

Mr. Ravichandar replied that they have not decided yet. It depends on what the Census Bureau wants.

Ms. Lockwood remarked that she used to work with Mr. Ambur at the U.S. Fish and Wildlife Service.

Mr. Ambur added that he wanted Oracle to speak about this project because it is a key marketplace player, it provides a good case study, and so that Oracle could give its through on XML querying.

Ms. Lockwood remarked that Oracle has many more projects in hold or planning stages. They are working on a proposal for Health and Human Services for an XML interface between its database and industry databases and the American Medical Association. They are beginning to see more and more agencies initiate XML projects.

Mr. Carmel asked why agencies are choosing to use a combination of XML and Oracle database models and not one or the other.

Ms. Lockwood replied that most agencies want dual functionality. Additionally, much depends on how the agencies are planning to implement XML.

Edward Fagbemi (Citrix) asked why the team chose to use graphical input devices (GIDs). He also wanted to know if they were used out of the box or if they were customized.

Mr. Ravichandar replied that if the common mechanism of exchanging information is XML, the GIDs can handle the exchange. In terms of accommodating Section 508 requirements, they have new requirements that ask the team to address this compliance. There needs to be some portable extension to the tool they have already designed.

Mr. Ravichandar then discussed some issues of accessibility as they pertained to wireless devices. There are several driving factors to accessibility, including anytime, anywhere access, application scalability to the Internet, extensibility and integration, and integrated business intelligence.

Key technology options for improving accessibility include Key Factor, Java, and XML.

Mr. Ravichandar then showed the WG a slide illustrating the components of a mobile application. The same application run on the wireless device should be able to run on a browser without any change in code. The components used in this demonstration are Oracle 9i and iFS.

iFS serves as a single place to store all of an organization's corporate information, an extensible file server based on Oracle8i, and a development platform that allows the integration of relational and non-relational data. iFS provides multiple interfaces.

Mr. Ravichandar used JDeveloper to develop the prototype demonstrated here.

Mr. Ravichandar then demonstrated how to access and search an XML document using a handheld device. Sample screenshots are available in the presentation.

Mr. Carmel asked if it was possible to work with tabular material on a handheld device.

Mr. Ravichandar replied that is was, depending on the style sheet.

Gregory Thornton (USCB) asked if iFS served as the referencing system.

Mr. Ravichandar replied that it is a natural extension of the database. The user points it to the file systems he wants it to read and it pulls data back into the database. It is a drag and drop application.

Mr. Ambur asked what the WG should be aware of in terms of emerging XML query languages relative to SQL.

Mr. Ravichandar referred the WG to http://technet.oracle.com for more information about XML Query language.

Ms. Harvey announced that the Washington Area's XML Users Group would be meeting the evening of June 20 at the American Geophysical Union. The subject of the night's discussion would be WML and XML topic maps.

Thomas McCullough (SI International) reminded everyone that the National Capital Region chapter of the Data Management Association will host the "Metatopia" conference September 20-21st in Gaithersburg, Maryland. For further information, please visit the conference website.

Mr. Ambur announced that he has been invited to meet with Fairfax County Government officials on June 27 to help identify XML data and schema to use with respect to the county's park system. The federal Open Source conference will be held October 2-5. Mr. Ambur is trying to stimulate discussion of XML at that conference.