Intelligent Enterprise

Better Insight for Business Decisions

RSS
Webcasts
Whitepapers
Subscribe
Home

March 28, 2002

The Semantic Web

A semantic Web will enable automated use of disparate, distributed Internet information sources and services

Seth Grimes

Tim Berners-Lee invented the World Wide Web in 1989 by creating a language for presenting and linking content, an information-interchange protocol, and basic client/server software. By 1994, the year he founded the World Wide Web Consortium (W3C) at the Massachusetts Institute of Technology, his attention had turned to embedding meaning in his creation.

This interest was long before the birth of well-loved sites (depending on your interests) such as SlashDot, The Onion, and Gap.com Applying business intelligence (BI) and knowledge management (KM) jargon, these kinds of sites are "stovepipe" systems, designed and used primarily for a single purpose, whether furthering technology development, commentary, or mass-market retail. Regardless of their purposes, these isolated systems constitute "islands of information." Much of the work in the BI and KM worlds is to build bridges among such islands. The thought is that a whole is greater than the sum of its parts, that acquiring new, varied data sources will expand the scope of analyses and strengthen their reliability. Fully integrating, rather than just occasionally drawing from those sources and finding relationships in the integrated whole, can bring additional rewards.

Building Bridges

In the BI world, developers build bridges using harmonized metadata and coordinated access methods. KM adds value by classifying and cataloging unstructured information to create analytic metadata. But, BI bridges are often one-directional and purpose-built, designed to support ad hoc queries and periodic reports, aggregation and consolidation via scorecards and portals, and electronic data interchange (EDI) extractions from specific sources to specific users. Generalized interchange, where information can flow in more than one direction and the user and source haven't negotiated protocols in advance, requires standardized, published interfaces with common metadata structures and definitions.

Interchange standards such as the extensible markup language (XML) and efforts such as the Dublin Core metadata project - which seeks to promote "the widespread adoption of interoperable metadata standards and developing specialized metadata vocabularies for describing resources that enable more intelligent information discovery systems" - are significant advances on the generalization road.

Searching for Meaning

Internet search engines compensate for the lack of meaning in the Web by analyzing sites for keywords and concepts that match user-supplied search terms. E-commerce exchanges attempt to meet integration and automation needs in the collaborative e-business arena - procurement, supply, logistics, and so forth - by mediating transactions via XML-encoded, domain-specific vocabulary and syntax. And, the latest much-hyped Internet hot ticket, Web services, takes a similar, server-centric approach, proposing universal description, discovery, and integration (UDDI) registries based on somewhat rigid definitions of what constitutes a service.

Search engines, exchanges, registries: The Web was supposed to be a disintermediated world, free of gatekeepers, where anyone can publish - "information wants to be free" - and agents automate routine tasks. The issue with these technologies is not only control, but also how best to accommodate and exploit the distributed, diverse, and constantly evolving nature of the Web.

These lofty concerns are not priorities for companies attempting to deliver particular products in a competitive marketplace. Despite the supposed dot-com bust, entrenched brands can still emulate Amazon.com by using bold applications of technologies that circumvent conventional production and delivery methods. Businesses often seek to protect their market positions by building fortresses, isolating themselves behind the walls of anticompetitive, proprietary platforms and business practices that together create barriers against innovative rivals.

This strategy was the shape of the Windows devolution, where technologies that support efficient, automated computing were (temporarily) displaced by interfaces that require laborious mouse work. The "personal" in personal computer is about enabling mass marketing of dumb systems, where the need for intelligence is shifted to users. Our PCs process information, but they don't understand it.

Search engines help circumvent walls and overcome isolation. Keyword search was a start. The next generation of engines examined context by automated or manual examination of links to and from pages and by translating keywords and summaries into concepts. Automated classification and categorization of results is now common; emerging search technology discerns attributes in pages and matches them to profiles that describe the properties of results to target.

The semantic Web, if realized, could support all these goals and more. The aim is to enable profile-based software agents not only to search but also to act on what they find. A semantic Web would shift the information processing burden back where it belongs, to a world of automated, peer-to-peer computing where people spend more time thinking about business and life problems and less time trying to remember whether "Preferences" is under File, Edit, or View on a program's menu bar.

Visualize a closed loop networked system that classifies and categorizes via KM and data mining techniques, applies algorithms to score and rank results, and derives and executes business rules. Achieving this ambitious synthesis will require creation of multidisciplinary standards that are independent of business domain and implementation technology. The W3C has been guiding this effort.

Raising the Standard

W3C participants have understood the need for a systematic but decentralized way to describe the contents and use of Web-hosted information that reflects the Internet's distributed nature. Although semantic Web standards have been carefully stewarded, competing and complementary Web services directory standards have recently and rapidly emerged. Some very predatory, domineering business interests are, of course, among their most prominent backers.