Metastructures 1998
&
XML Developers' Conference

The Graphic Communications Association (GCA) invites you to participate in the fifth Metastructures conference (formerly known as the International HyTime Conference) and the third XML Developers' Conference (formerly known as XML Developers' Day).

The Topic Navigation Map formalism is revolutionizing the sharing of knowledge, the representation of corporate memory, and the provision of convenient, orderly access to all significant aspects of otherwise intractably chaotic and diverse information resources, including read-only resources. The topic map formalism was originally developed using SGML and HyTime linking/addressing, but useful topic maps can also be created using XML, Extended XLink and XPointers.

This tutorial will provide participants with a useful working knowledge of topic maps. The topic map architecture will be explained and demonstrated. The class will work through examples interactively, and demonstrated using the EnLightEn topic map technology and methodology. The benefits of the paradigm to end users will be demonstrated using an ordinary Web browser.

One of the fundamental notions of SGML is that many documents can all conform to a single model. For example, probably the single model to which the most documents conform today is called "HTML". The model usually used for HTML is some version of the "HTML DTD" (the "HyperText Markup Language Document Type Definition").

Today's leading-edge, productivity-enhancing information management systems are increasingly relying on a significant refinement of the "one single model" idea: a given document can now conform to more than one model simultaneously. Such multi-model documents can be viewed and used with a variety of systems, and they can convey a wide variety of meanings, using a wide variety of structural conventions, each of which is already widely understood and popularly implemented. This paradigm is variously called "Architectural Forms (AFs)", "Architectural Form Definition Requirements (AFDR)", "SGML Architectures", and sometimes "Inheritable Information Architectures." Software tools that are in the public domain already support it. The AF paradigm is just as useful with XML as it is with SGML.

This tutorial has been offered by the instructor at major information management sites, and it has led to significant economic commitments to the paradigm. Participants will gain the knowledge and skills necessary to borrow information constructs from existing DTDs, to use them successfully as metastructures in their own information architectures, and, using a freely available tool, to verify automatically that documents that claim to use such borrowed constructs actually do structurally conform to each and every one of them, thus helping to guarantee reliable information interchange.

If you need practical experience with XLink/XLL and XPointers, this is the tutorial you want.

The "HyTime" standard (the Second Edition of which was published by the ISO in 1997) is like a large grab-bag full of metastructures and metastructural techniques and tools. Some of these, like the "varlink" metastructure, have already been borrowed for other standards ("XLL", aka "XLink"). Others, like "valueref", seem likely to follow. Two of the metastructural techniques standardized by HyTime, known respectively as "architectural forms" and "groves", are the topics of other tutorials on this same program. HyTime is unique in its holistic approach to the problem of metastructures and information management. HyTime's metastructures are designed to be compatible with each other, to take advantage of each other, to take advantage of certain classes of non-HyTime metastructures, and not to interfere with an information architect's creativity.

This tutorial, a very popular tradition at these conferences, provides a guided tour of HyTime, and gives participants useful knowledge about metastructures in general, and about specific metastructures and metastructural techniques. This year, special emphasis will be placed on XLink/XLL and XPointers, the W3C draft linking metastructures for XML. There will be interactive examples and demonstrations.

XML is the document structuring and markup language for everyman. With much of the representational power and flexibility of its ISO-standard forebear, SGML, XML is leaner, easier to learn, easier to implement, and still phenomenally flexible and powerful. XML has hardly even begun to revolutionize the World Wide Web, but it is widely acknowledged that such a revolution is inevitable and has in fact already started. XML brings limitless possible information architectures to a world where, only recently, there was only one possible architecture, called HTML. The potential for practically every kind of business to exploit Web communications in self-determined and community-determined ways is staggering, but at this early stage, practically unknowable.

This tutorial will acquaint participants with the basic features of XML and how XML is used. Participants will gain an understanding of XML sufficient to spark their own creativity in the burgeoning field of Web communications, an appreciation of the fabulous potential of XML for solving business communications problems, and they will see XML at work.

The "grove/property set" concept is the theoretical basis of the state of the art in information component addressing, formatting, and re-use. In the case of an XML or SGML document, the parser produces a "grove" as the result of the parsing process. The grove is a set of objects ("nodes") bearing certain relationships to one another, expressed as "property values". There is an ISO standard schema ("property set") for SGML; the nodes in an SGML grove conform to the classes defined in the SGML property set. The property set for XML will likely resemble both the SGML property set and the DOM. There can also be property sets for specific XML or SGML information architectures, such as XLL and HyTime. In the case of an XLL document, an XLL engine produces an additional XLL grove in which the semantic properties of the XLinks appear as a convenient API. Many notations other than XML and SGML, including graphic notations and word-processor file formats, are also amenable to the grove paradigm, which makes possible a single holistic approach to addressing practically everything, and which explains why the HyTime standard is implementable in such a way as to allow HyTime documents to reference, link to, and reuse anything, anywhere, in any convenient terms. A minimal usable property set definition for CGM graphics has been demonstrated, and a more comprehensive one is in development.

Participants should be familiar with SGML or XML syntax, be able to read a DTD, have some grasp of object orientation, and understand how an SGML or XML document instance can be regarded as a tree structure.

Participants will gain an understanding of the grove concept, and the aspects of the SGML property set that are equally applicable to XML. After the tutorial, they will be able to delve further into the details of property sets with confidence. The emphasis will be on the features of SGML/XML groves, rather than on the HyTime/DSSSL/XLL applications of the grove paradigm. There will be just enough description of applications to illustrate the essential concepts.

Conference Registration Desk opens for business

Sam Hunting (independent consultant)
"Architectural Forms in Legal Contracts"

Legal agreements are ubiquitous and powerful documents that exhibit endless varations on common themes. In complex machine-mediated transactions, such as may occur on the Web, contracts need to express the relationships and contingencies they establish in a machine-readable fashion. SGML and XML information architectures for contracts will be needed, and architectural forms can be used as metastructures that will provide both flexibility and unambiguously interchangeable semantics. Following both the modern "programming by contract" movement and 19th Century formalisms in contract law, the essential components of a contract are found to be: client, supplier, preconditions, postconditions and invariants. For the sake of simplicity, these components are developed and elaborated using a Property Set. Potential applications include electronic commerce, workflow, and the potential disintermediation of the legal community from routine transactions. Limitations and extensions of the formalism are discussed. The appropriate forum for development of a simple contract architecture would be transnational and non-profit.

Jennifer Yang and Grant Taylor (Digital Graffiti)
"Media Asset Management Software for Graphic Artists"

In designing media asset management software for graphic artists, there were many advantages to using recognized standards such as the Anglo-American Cataloguing Rules and XML. The purpose of the software is to track physical pieces of art and computer files on an equal footing -- in other words, to maintain information regarding both physical and soft assets -- so that all resources are supported by a single powerful finding aid. Importing and exporting the metadata as XML allows users to share their metadata across platforms and over the web. This enables graphic artists, for example, to promote their work and to provide catalogues.

Paul Prescod (ISOGEN International)
"Forest Automata: a formalism for document schemas and validation" [Or, Formalizing SGML and XML Instances and Schemata with Forest Automata Theory]

The characteristics of the tree-like element containment structure of SGML and XML documents can be constrained by schemas called Document Type Definitions (DTDs); each schema describes a class of documents. The declarations found in DTDs can be viewed as an extended context free grammar (CFG) for a simple class of language. Unlike the syntax of full-blown SGML, the syntax of XML is simple enough to allow XML documents to be parsed without knowing the grammar. Instead, we can look at the grammar as a form of constraint not on the document string, but on the tree that results from that string. We can describe the tree in terms of the forest automata theory, which describes classes of forests (forest languages) instead of strings. The new formalism is a generalization of string automata theory, and it supports concepts such as forest regular languages, forest regular expressions and forest regular grammars. Given union, concatenation, linear tree homomorphisms and other, similar operations, we can build a full query and transformation language.

Note: A provisional document is available online: "Formalizing SGML and XML Instances and Schemata with Forest Automata Theory."

(break)

Neill A. Kipp (Virginia Tech)
"Flying Saucers, Wizards, and Deep Space:
HyJinks in the Virginia Tech Digital Library Research Laboratory"

Virginia Tech's Digital Library Research Laboratory has developed a formalism for digital libraries called "Streams, Spaces, Structures, and Scenarios (4S)", a new electronic thesis and dissertation markup language (ETD-ML), and an online Speculative Fiction library (VTSF) that uses HyTime finite coordinate space metastructures for anchor addressing (http://video.dlib.vt.edu:90/saucer/).

But wait, there's much more! The HyperWizard electronic thesis and dissertation (ETD) submission software "wizard" allows ETD authors to provide metadata. An XML document type definition (DTD) whose elements resemble those found in the Metafile for Interactive Documents (MID) proposed standard, and whose formatting software will handle all CGI interactions, allows XML document authors to write CGI wizards without becoming Perl experts. The "HyperWizard" document type has the usual title and paragraph elements, but it also includes user interaction metastructures: button, textfield, script, variable, if, else, function, while, and goto elements (goto being implemented as a HyTime hylink). Because the Wizard is document-driven, it works as Perl/CGI, but it also can run as a Java Applet.

Soon, ETDs will be delivered to users of the Networked Digital Library of Theses and Dissertations (NDLTD) using a spatial retrieval metaphor. A delivery engine for three-dimensional finite coordinate spaces (FCSs) creates virtual environments of the rooms in a traditional library. Using this tool, we plan to deliver ETDs from the NDLTD to Web clients. The library is divided into floors and rooms based on a controlled vocabulary derived from college and department information. Flying through the collection should give users an immediate feeling of the breadth and depth of the collection. "Help" and "search" are provided within the metaphor by means of reference desks and hyperlinks to information pages and search dialogs. The ETD collection items can be viewed "media-first": the pictures in the collection appears as art on the "walls"; each picture is hyperlinked to its corresponding ETD.

Eric W. Johnson (Neville & Associates)
"Non-musical Applications for Hypermedia/Time-based Structuring Language"

This brief but pivotal paper was originally contributed in 1990 to the ANSI X3V1.8M committee that was designing the Standard Music Description Language (SMDL). The user scenarios it describes influenced the transformation of the SMDL draft standard into what eventually became the HyTime standard.

(lunch)

David M. Price (IBM Advanced Manufacturing Solutions)
"The STEP Data Integration Architecture Activity"

Standardized data models for products and process have been developed in the context of the ISO TC184/SC4 Industrial Data subcommittee, the primary product of which is ISO 10303, the STEP Standard for the Exchange of Product model data. The benefits of exchanging design engineering data between applications using STEP are now being realized. However, more data interoperability, higher levels of enterprise integration, and more convenient data sharing are needed. Product information expressed using metastructures other than STEP Application Protocols, as well as many kinds of information that are not product information, must be integrated and shared by multiple applications and users concurrently. The models of such information may or may not be defined using STEP's EXPRESS language.

Standards-based capabilities for exchange and integration of industrial data can be extended without compromising existing investments in STEP, PLIB, SGML, etc. A data integration architecture can support a broad set of requirements without replacing the underlying metastructures of existing data and applications.

Philippe Futtersack (Electricité de France)
"Modeling, Generation, Storage and Presentation on the Web of One Million HyTime Links"

The Electronic Library Project was undertaken by the R&D Division of Electricite de France, the French national electric utility. The R&D Division of this giant nuclear power company is large, heterogeneous, and dynamic, and it generates many technical documents. The purpose of the Electronic Library Project is to describe all ongoing activities within the R&D Division, and to explore how ODBMS, SGML/HyTime, and advanced user interface technologies can provide efficient access to technical information. This paper focuses on the creation, management, and use of hyperlink metastructures, and shows that the new techologies can give very good results in the context of a large research organization.

David Williams (Woodward Governor Company)
"Annotation of Engineering Documentation"

Engineering documentation comes in many flavors, is time sensitive, dynamic, and long-lived. Those who use these documents need to be sure that they have the latest information and are made aware of changes as quickly as possible. The creation and maintenance of aircraft maintenance manuals, for example, is complex, and alterations to the authoritative data can only be made according to highly constrained processes.

A system that would enable engineering documentation to be annotated on-the-fly in a quality-conscious, interactive, controlled way, would allow the latest data to always be made available and also provide a means for automated change. The use of extended (aka "independent", "out-of-line") hyperlinking metastructures will improve productivity and overall product quality.

(break)

Francois Chahuneau (AIS, Berger-Levrault)
Steven R. Newcomb (TechnoTeacher)
"SGML and Schemas"

A perspective on the relationship between SGML and data modeling concerns. A plea that APIs to structured data are better modeled separately from their interchange architectures. A hope that the real correspondences between interchange architectures and their most convenient APIs can be expressed declaratively.

Nick Ruddock and Gerry Berger (Mincom Pty Ltd)
"A Practical Double Blind Test of the HyTime Model"

A good theoretical model is "characterized by the fact that it makes a number of predictions that could be disproved or falsified by observation. Each time new experiments are observed to agree with the predictions the theory survives, and our confidence in it is increased" (Stephen Hawking in "A Brief History of Time", citing Karl Popper).

LinkOne was originally developed as a commercial software package to meet the practical needs of Union Pacific Railroad for publishing, distributing and viewing heavy equipment parts catalogues in electronic form. The specification also included the requirement to order parts, through their business system, directly from the parts catalogues. LinkOne's custom relationship model handles a large number of information distribution channels each supporting a different view of the information and the framework within which it resides. To make the information maintainable, there is a method for expressing variable information content and relationships between external information sets.

Recent studies have demonstrated that the HyTime metastructures can be very effective in expressing the interrelationships required for variable content and multipurpose LinkOne information sets. Therefore the development of LinkOne can be regarded as an experiment with results that support the underlying theoretical model of HyTime hyperlinking, despite the fact that LinkOne and HyTime were developed quite independently of each other.

[January 18, 1999] "A Practical Double Blind Test of the HyTime Model." A paper by Gerry Berger and Nick Ruddock, presented to delegates at Metastructures 1998 & XML Developers Conference, Montreal, Quebec, Canada. August 18th, 1998. "When comparatively analysed, LinkOne displays fundamental concepts very similar to those defined in the ISO Standard called HyTime. This Standard defines a theoretical model that should be capable of describing very intricate relationships in complex documentation systems. LinkOne is a commercial documentation product, which, while it was developed independently of HyTime, embodies many of the key concepts found in HyTime and, as such, demonstrates that the theoretical model in HyTime can be effectively applied to real documentation problems. This paper recognises that LinkOne can claim to be HyTime compliant with respect to the fundamental concepts, and details the maintenance documentation system for heavy, repairable equipment. Further, the paper demonstrates how LinkOne is able to meet business goals using concepts that closely mirror those from the HyTime Standard."

W. Eliot Kimber (ISOGEN International)
"HyTime in the Fast Lane"

With the beginning of the public availability of the source code of the presenter's own PHyLIS system last May, grove-based systems offer unprecedentedly demonstrable assurance to information asset owners that their assets can be re-used and exploited in arbitrary combinations and contexts. The implications of this, and the experiences of the past year will be discussed by one of the foremost HyTime systems implementers.

(adjourn)

Conference Registration Desk opens for business

Yoshihisa Gonno (Information Broadcasting Laboratories)
Fumihiko Nishio
Kazuo Haraoka
Yasuaki Yamagishi
"Metadata Structuring of Audiovisual Data Streams on MPEG-2 System"

MPEG-2, the standard generic coding for moving pictures and associated audio information (including TV programs), is now gaining worldwide popularity. MPEG-2 systems provide high speed transport protocols optimized for compression and transportation of audiovisual data streams on digital broadcasting systems. MPEG-2's efficient use of bandwidth, which allows hundreds of simultaneous transmissions to be delivered through a single broadcasting satellite to millions of set-top boxes, creates a situation in which viewers need improved navigation aids in order to cope with the flood of services.

DVB-SI is a proposed metadata description framework for transporting information about audiovisual contents to users of MPEG-2 systems. The framework includes a so-called electronic program guide (EPG) and to select specified contents from broadcast data streams (see next presentation). However, DVB-SI does not provide metastructures for describing information components within the selected content streams. In order to provide sophisticated multimedia services with continuous audiovisual data streams, subparts of a stream must be addressable so they can be restructured for individual users in specifiable ways. XML and SMIL syntax can be used in a framework of metadata description that presents the possibility of unifying broadcast and web production processes. As an extension of DVB-SI, a framework of metadata transportation for converting XML structured metadata and a metadata schema representing an XML namespace, separately into the DVB-SI section format, can provide high speed and lightweight metadata transport mechanisms optimized for broadcast-based protocols.

Fumihiko Nishio
Yoshihisa Gonno (Information Broadcasting Laboratories)
Kazuo Haraoka
Yasuaki Yamagishi
"Transporting RDF Metadata Associated with Structured Contents"

The Electronic Program Guide (EPG) feature of MPEG-2 systems (see previous presentation) is widely used to select TV programs, but it provides merely timeline-based guidance. In future digital broadcasting systems, metadata will be used for content filtering and content navigation based on the personal preferences of individual viewers. An efficient filtering capability, based on simple data structures, is needed. Richer semantic descriptions are also required for efficient navigation of highly structured content. The Resource Description Framework (RDF) provides a simple data model for describing the semantics of resources.

We utilize RDF's metadata description capability to represent metastructured content, transporting RDF metadata via the MPEG-2 system's section format. The encoded metadata optimizes transmission and supports the full metadata description power of RDF. Thus, both efficient content filtering and flexible content navigation will be realized by using RDF metadata to characterize MPEG-2 contents.

Jonathan Robie (Texcel Research)
"XQL (XML Query Language)"

XQL (XML Query Language) is a query language designed specifically for XML documents. In XQL, the structure and content of XML documents is the logical model for queries, addressing, and patterns.

XQL provides a way to specify nodes, paths, and conditions that is based on an model of logical completeness for the structure of XML documents. The format of query strings is compact, easy to type and read, and easily embedded in programs, scripts, URLs, and XML or HTML attributes. XQL was designed to support naive implementations, but also to allow query optimization in more sophisticated implementations. In a query environment, XQL operates on XML and returns XML as output. When tree reshaping is needed, XQL can be combined with XSL.

(break)

Bryan Bell (Frank Russell Company)
Michel Biezunski (High Text SARL)
"Topic Maps and the Information Age"

As the benefits of the Web's worldwide interconnectedness are realized, there is growing awareness of the difficulty of providing useful and efficient access to massive amounts of information. Individuals and organizations are looking for better ways of developing, managing, and navigating heterogeneous repositories around the globe. The Topic Navigation Map architecture, first introduced as a HyTime application, is nearing maturity with the approaching publication of its specification, ISO/IEC 13250, and is fully supportable in XML using XLL. This elegant package of metastructures, the result of years of work by many experts, is being applied by several large organizations where efficient access to hard-to-find items within enormous heterogeneous information bases is a strategic issue of the utmost importance.

Dianne Kennedy (SGML Resource Center)
"XML-Data Update"

XML-Data is a new schema language based on XML. Handy and pragmatic, it includes features that, taken together, make XML-Data a unique tool for expressing constraints on the syntactic, conceptual, and relational characteristics of information represented either in XML or in any other manner. For example, XML DTDs can be expressed in XML using XML-Data, as can the row types key relationships common in relational data. XML-Data is already being used in the automobile manufacturing industry.

(lunch)

Murray Maloney (VEO Systems)
"Electronic Commerce and the Common Business Language"

XML will enable businesses to preserve the richness of their original database schemas when they publish a product or services catalog on the Web. XML is thus a key part of making Web sites and services "smart enough" to enable agents and arbitrary applications to extract information from them, making them function as "objects" in component-based commerce.

The eCo system is an object-oriented interoperability framework in which all services and resources on the Web are treated as business objects that can be combined as "plug and play" components to build virtual companies, markets, and trading communities. XML is a central part of the eCo architecture and is used to encode its "Common Business Language" that enables semantic interoperation and integration of different commerce applications.

Eve Maler (ArborText)
"XLL Technical Update"

The XML family of metastructures includes a draft hyperlinking and addressing component called XLL ("XML Linking Language", also known as, "XLink", "Simple Link" and "Extended Link"). XLL encompasses the XLink mechanism for strongly typed hyperlinks that can support the initiation of traversal from any anchor, as well as the XPointer language for addressing into structured XML documents. This presentation, by one of the central figures in the development of XLL, will provide the latest information on XLink, XPointer, and related developments.

Eric J. Miller (Online Computer Library Center [OCLC])
"RDF Technical Update"

Recent developments in XML's Resource Description Framework (RDF) are discussed by one of the editors of the relevant W3C technical document. The evolving RDF package of metastructures for metadata representation is set to become a lynchpin of Web-based information interchange and accessibility.

(break)

Neel Sundaresan (IBM Almaden Research Center)
"Summarizing Web Resources using SumML"

Web resources occur in various data forms - simple text, marked up text, binaries, multi-media streams, computer programs, container structures, and databases. Search engines require that these heterogeneous structures be described.

RDF (Resource Description Framework) is a graph language with an XML serialization syntax that provides metastructures for describing web resources such as PICS, P3P, Sitemaps etc. RDF defines resources, properties, and higher level structures, and it provides constraints, relationships, types, and models for resource descriptions.

SumML is a set of schemas for categorizing and summarizing web data that occur in various forms and can be intermingled in different ways. SumML uses RDF, supports algebraic, composing, customizing and specializing operations, and can be used to project on schemas and descriptions. SumML's use in a large-scale web project will be discussed.

W. Eliot Kimber (ISOGEN International)
"STEP/EXPRESS as XML; XML as STEP/EXPRESS"

STEP is a family of ISO standard metastructures for product description. The benefits to be realized by using STEP in combination with other metastructures are stunning: XML (or SGML) can greatly facilitate the interchange of STEP-conforming product information. STEP's powerful EXPRESS schema language offers a well-developed constraint formalism that can aid the development of a standard Activity Policy Expression Language. The Property Set formalism of the SGML Extended Facilities can be combined with STEP's corresponding formalism to provide more modeling power to both the STEP and SGML arenas. This paper discusses technical aspects of the ongoing harmonization work between STEP and the SGML and XML families of standards.

Daniel Rivers-Moore (Rivcom)
"An XML Representation of EXPRESS-driven data"

Existing STEP-defined metastructures for the interchange of STEP data in a vendor-neutral fashion are not optimized for widespread use in a marketplace where data interchange is increasingly accomplished using XML. STEP data can, of course, be interchanged using XML, given an appropriate XML-based information architecture that is widely understood and accepted. The implications of using XML for interchanging STEP data, the potential impacts of EXPRESS-modeled data flowing through XML channels, and the relationships between EXPRESS, XML-Data, RDF, etc. are discussed.

[January 18, 1999] See now: "XML and EXPRESS as Schema Definition Languages." By Daniel Rivers-Moore [Director of New Technologies, RivCom, and Joint Project Leader of the ISO Preliminary Work Item on SGML and Industrial Data]. Paper presented at the MetaStructures 1998 Conference (August 17 - 19, 1998. Le Centre Sheraton Hotel, Montréal, Québec, Canada). Abstract: "EXPRESS is a rich and mature language for the definition of data schemas. It is part of the STEP standard (ISO 10303), and is in widespread use to define data models for large-scale industrial applications, including manufacturing, engineering, defence, oil rigs, processing plants, etc. As part of the ongoing STEP/SGML harmonization activity under ISO TC184/SC4, a New Work Item proposal has just been put forward under the title 'An XML representation of EXPRESS-driven data'. The scope of the work item includes the use of XML to encode both the EXPRESS schema definition, and the instance data that conforms to that schema. This presentation will outline the main semantic features of the EXPRESS language, and show the current state of work on defining an XML representation of those semantics. The relationship of this work to other ongoing efforts to develop an XML syntax for schema definitions will be discussed." See also: "SGML and STEP."

(adjourn)