Now It's Personal

Search is an important factor in making portals the personalized knowledge gateways they should be

By Carl Frappaolo & Hadley Reynolds

In a relatively short time - fewer than 2,000 days, so far - the World Wide Web has changed our relationship to information and the very underpinnings of our personal and professional lives forever. We are both eyewitnesses to, and laboratory animals for, an epochal transformation from information scarcity to information surplus.

Today the Web warehouses one billion public pages. The "deep Web" - that part of the network beyond the realm of the conventional search engines - comprises dynamic pages created and personalized on the fly and other transitory content amounting to another 50 billion pages. Public Web sites number nearly 40 million today, a figure that grows by nearly 50,000 every day. These numbers are testimony enough to the immense production capacity of human knowledge, but the most compelling fact about this collection is that it continues to double in size annually.

Arrayed against this tsunami of e-information are what we generically call "search" technologies. But to prevent a flood of Web content from swamping e-business strategies and organizational knowledge processing, managers are now looking hard at new, potentially more effective approaches to search.

Once viewed as the domain of corporate librarians and professional researchers, the Web revolution has recently redefined search as a fundamental and newly mission-critical technology: In essence, search is the intersection of integration, categorization, linguistic analysis, and Web-context analysis. To survive in the "age of e," land-based and e-businesses alike need to take a new look at its role and requirements.

Categorization: Info-Navigation Foundation

Although most people wouldn't consider it a search tool, categorization is in fact the foundation for an effective information experience in either an e-business or knowledge portal environment. Its primary role is to build information context that lets business professionals focus on the actionable essence of their e-environment.

Within each organization, elements such as current operations practices; management initiatives; corporate history, structure, and culture; and available professional resources and learning requirements create a context for working with information. Successful portal designs effectively reflect this "knowledge map" in their category structure - usually, by supporting categorization at the user, department, corporation, process, product, customer, and partner levels.

There are four common approaches to categorization: automatic, manual, dynamic, and modifiable. Portals often implement a combination of them:

• Automatic categorization relies on software tools that assess information collections and repositories to extract taxonomic characteristics and hierarchies. These tools rely on statistical or semantic methodologies. Automatic tools can help create categories for information collections that are not tagged.
• Manual categorization relies on architects, analysts, and users to analyze corporate information and identify useful categories and hierarchies. Maintaining such hierarchies is often a manual process. Although this method is accurate and tailored to the organization, it is labor intensive and a challenge to maintain.
• Dynamic categorization is a type of automatic approach that maintains and updates the information hierarchy dynamically based on additions or deletions to the repositories and changes in the organization's processes and use of information. Dynamic tools can potentially make developing and maintaining your portal easier.
• The variety of business operations in any given enterprise often requires different categorization taxonomies for different departments. In addition, the rapid pace of business "Internet time" may require you to give users the ability to change existing categories through modifiable categorization. This capability yields improved customer intimacy, greater responsiveness to changing markets, and enhanced knowledge management practices.

With the increasing volume of potentially relevant information online, the ongoing maintenance challenge of categorizing new documents as they arrive is a killing task for manual approaches on public or corporate portals. However, automated categorization and hyperlinking technology is beginning to come online - technology that can greatly help portal implementers create and support ongoing classification and linking.

Tomorrow's extensible markup language (XML)-enabled information applications will add a built-in classification element that will make categorization substantially easier. With the current capabilities of today's technology, however, categorization in the corporate portal will continue to call for some level of human intervention, informed by professional library science, with assistance from automated software tools.