- Information Is Interesting Stuff
- Information and Structure Are Inseparable
- Formal Languages Are Easier to Compute Than Natural Languages
- Generic Markup Makes Natural Languages More Formal
- A Brief History of the Topic Maps Paradigm
- Data and Metadata: The Resource-Centric View
- Subjects and Data: The Subject-Centric View
- Understanding Sophisticated Markup Vocabularies
- The Topic Maps Attitude
A Brief History of the Topic Maps Paradigm
The work on topic maps began in 1991 when the Davenport Group was founded by UNIX system vendors (and others, including the publisher O'Reilly & Associates). The vendors were under customer pressure to improve consistency in their printed documentation. There was concern about the inconsistent use of terms in the documentation of systems and in published books on the same subjects. System vendors wished to include O'Reilly's independently created documentation on X-Windows, under license, seamlessly in their system manuals. One major problem was how to provide master indexes for independently maintained, constantly changing technical documentation aggregated into system manual sets by the vendors of such systems. The first attempt at a solution to the problem was humorously called SOFABED (Standard Open Formal Architecture for Browsable Electronic Documents).
The problem of providing living master indexes was so fascinating that, in 1993, a new group was created, the Conventions for the Application of HyTime (CApH) group, which would apply the sophisticated hypertext facilities of the ISO 10744 HyTime standard. HyTime had been published in 1992 to provide SGML with multimedia and hyperlinking features. The CApH activity was hosted by the Graphic Communications Association Research Institute (GCARI, now called IDEAlliance). After an extensive review of the possibilities offered by extended hyperlink navigation, the CApH group elaborated the SOFABED model as topic maps. By 1995, the model was mature enough to be accepted by the ISO/JTC1/SC18/WG8 working group as a "new work item"a basis for a new international standard. The topic maps specification was ultimately published as ISO/IEC 13250:2000.8
During the initial phase, the ISO/IEC 13250 model consisted of two constructs: (1) topics and (2) relationships between topics (later to be called associations). As the project developed, the need for a supplementary construct, one able to handle filtering based on domain, language, security, and version, emerged; as a result, a mechanism for filtering was added, called facet. This approach was soon replaced by a more powerful and elegant vision based on the notion of scoping. The notion of scope in topic maps is one of the key distinguishing features of the topic maps paradigm; scope makes it possible for topic maps to incorporate diverse world views, diverse languages, and diversity in general, without loss of usefulness to specific users in specific contexts and with no danger of irreducible "infoglut."
As an aside,9 note that the scope and subject identity point aspects of the topic maps paradigm were first developed and articulated by Peter J. Newcomb and Victoria T. Newcomb during a 1997 breakfast conversation at the Whataburger restaurant in Plano, Texas. In our family, we still sometimes call those aspects the Whataburger model, although the Whataburger interchange syntax has not survived. The XTM conceptual model accurately reflects the Whataburger model, however; it has stood the test of time. It's interesting to note how the syntax of topic maps has evolved since Whataburger. The syntax that minimally and accurately reflected the Whataburger model turned out to be inexplicable to most people; it was a marketing fiasco. Michel Biezunski, who for many reasons is the primary hero of the story of topic maps, is not coincidentally also the origin of what I call Biezunski's Principle. Simply put, Biezunski's Principle is: There is no point in creating a standard that nobody can understand. (Another way he sometimes puts it is, "I'm not interested in convincing anyone that we are smarter than they are.") The whole idea of having a syntactic element type that corresponds to the notion of a topic is, in strictly technical terms, totally unnecessary baggage that actually obscures the deeper and beautifully simple structures that topic maps embody. Even so, the <topic> element type is the foundation of the syntax of topic maps, both in the ISO standard and in the XTM specification. This is because people intuitively and quickly grasp the notion of <topic> elements, and the whole idea that a topic can be represented syntactically as a kind of hyperlink is an inherently exciting one. For me, the popularity of the <topic> element type and the marketing success that the topic maps paradigm now represents are convincing demonstrations of the power of Biezunski's Principle. (I think Biezunski's Principle owes much to the work of Tim Berners-Lee and others, whose design for the World Wide Web succeeded in opening a whole frontier of human interaction and endeavor, where other designs, including more intellectually elegant and powerful ones, had failed to get serious global traction. But that's another story.)
The ISO 13250 standard was finalized in 1999 and published in January 2000. The syntax of ISO topic maps is at the same time very open and rigorously constrained, by virtue of the fact that the syntax is expressed as a set of architectural forms.10 (Architectural forms are structured element templates; this templating facility is the subject of ISO/IEC 10744:1997 Annex A.3.11) Applications of ISO 13250 can freely subclass the element types provided by the element type definitions in the standard syntax, and they can freely rename the element type names, attribute names, and so on. Thus, ISO 13250 meets the requirements of publishers and other high-power users for the management of their source codes for finding information assets.
However, the advent of XML and XML's acceptance as the Web's lingua franca for communication between document-driven and database-driven information systems created a need for a less flexible, less daunting syntax for Web-centric applications and users. This goal, which was achieved without losing any of the expressive or federating power that the topic maps paradigm provides to topic map authors and users, is the purpose of the XTM (XML topic maps) specification.
The XTM initiative began as soon as the ISO 13250 topic maps specification was published. An independent organization called TopicMaps.Org,12 hosted by IDEAlliance, was founded for the purpose of creating and publishing an XTM 1.0 specification as quickly as possible. In less than one year, TopicMaps.Org was chartered and the core of the XTM 1.0 specification was delivered at the XML 2000 conference in Washington, DC, on December 4, 2000, with the final version of XTM 1.0 delivered on March 2, 2001.
Michel Biezunski (of InfoLoom) and I (of Coolheads Consulting) were the founding cochairs of TopicMaps.Org and coeditors of the Core Deliverables portion of the XTM specification as well as of the remaining portions of the Authoring Group Review version of the specification. In January 2001, Graham Moore (of Empolis) and Steve Pepper (of Ontopia) became the new coeditors, and Eric Freese (of ISOGEN/DataChannel) became the chair of TopicMaps.Org. More recent events in the history of XTM and TopicMaps.Org are discussed in Chapter 4.