Vote Now for the KMWorld Readers' Choice Awards !

Representing knowledge in enterprise portals

This article appears in the issue May 2001 [Volume 10, Issue 5]

Portals, and portal technology, caught the imagination of IT users and vendors in the late 1990s. Enterprise portals hold out the promise of knowledge aggregation and integration across an entire organization. They represent one of the holy grails of contemporary knowledge management: Portals permit an organization to store and retrieve information for activities like problem solving, dynamic learning and strategic planning. Portals are important because they organize corporate knowledge and provide a rich collaborative environment.

Furthermore, power in KM organizations is often decentralized, and many users are granted responsibility for corporate decision making. Those workers want to access data themselves. Employees require quick, on-demand access to corporate information.

In this context, mission-critical data must be presented in a way that facilitates insight and fosters informal decision making. Corporations have responded to the challenge by deploying Yahoo-style directories and animated taxonomies in their corporate portals. The value of information is increasingly tied to interpretation, and these content management tools help users make sense of enterprise data.

Most PCs and networks utilize specialized, task-based applications such as word processing or spreadsheet programs, but it's difficult for workers to see connections between related fields of knowledge. Document classification and data visualization tools make enterprise knowledge easier to navigate and understand.

Information structures, such as topical directories and animated taxonomies, help workers find the information they need. They integrate data from disparate sources and illustrate the conceptual relationships that exist within and between topics. For example, browsing a hierarchy devoted to HR information is a good way for employees to learn about benefits, education opportunities, internal job openings, etc. If a worker wants to research company-sponsored awards, this taxonomy would serve as a logical starting point.

Knowledge organization software can be divided into two types: automatic classification technology and data visualization tools. Some vendors like Semio sell products that automatically funnel documents into hierarchical directories. This technology creates Yahoo-style directories with minimal human intervention. By contrast, animated taxonomies, sold by Inxight, The Brain and ThinkMapk, graphically represent large amounts of data. This software translates enterprise knowledge into an animated tree or web structure. In a visually stunning interface, wire frame graphics link a category with all its sub-categories.

While both automatic classification software and animated taxonomies share the goal of easing information access, they employ different techniques of information organization. One important difference between classification and visualization tools is that animated taxonomies do not materially reorder content. Yahoo-style directories physically catalog documents in an enterprise portal. In contrast, visualization tools graphically link similar content together regardless of where material is located. Animated taxonomies are navigational devices (as opposed to cataloging tools) that promote knowledge retrieval.

What's unique about a digital environment? Before examining the topical directories and animated taxonomies in detail, let’s step back and consider the special qualities of the Web as an information medium. Sophisticated portal design means acknowledging both the limitations and benefits of digital applications and platforms. There are three distinct ways that people use the Internet:

  • as a place where tasks are performed (e.g., online banking, registering for membership in a professional organization, writing a proposal with a group of people or conducting research in a specialized database);;

  • as a document delivery medium where information is read off the screen, printed out or saved to disk (e.g., posting company news on an intranet, accessing The New York Times, sending e-mail);;

  • as a point of direct experience, a platform for fun and entertainment [e.g., digital art Web sites such as "Modernism" at The Minneapolis Museum of Art or the official Motown records Web site.;

Corporate portals will most likely utilize the Web according to the first two methods listed above. In a task or document delivery environment, revealing the inherent structure of information or mapping topics into a classification scheme is very important. Information structure can be prescriptive, in which organization is imposed on a body of texts (e.g., any manually constructed taxonomy), or descriptive, in which categories that "naturally" occur within the corpus are chunked together automatically. Navigational structures are essential to corporate portals because they help users interpret data and identify meaningful patterns in a body of texts.

Directories and information access Automatic classification software funnels documents into a hierarchical structure composed of parent/child relationships. These directories can be hand-crafted or automatically generated from the text itself. The hierarchies are hand-built and then automatically populated with content. On the other hand, Semio . for example, creates the top level of a taxonomy manually or employs pre-determined thesauri, but the lower levels of the directory are automatically generated from textual content.

Directories provide a meaningful context for retrieved information because they delineate conceptual relationships. Bitpipe, a vertical portal that syndicates IT information, is an excellent example of an online directory. Within the category "Databases," searchers can hop from one associated concept to another (from "Reporting Software" to "Transaction Processing"), learn about related terms ("Database Management" and "Markup Language"), or begin their search at a broader term in the hierarchy and move down to more specific instances of a concept (begin at the top of the "Enterprise Applications" hierarchy and browse down to "Databases").

Searching is an iterative venture, and people often can't fully articulate what they're looking for. This type of information structure takes the pressure off users. Directories contextualize the search and knowledge management process. Because classification schemes explicitly delineate a structure of conceptual relationships, the meaning of a discrete term is not presented as "a thing in itself" but is understood in the context of surrounding terms.

Animated taxonomies Animated taxonomies transform a body of texts into swirling, dancing taxonomies. Inxight, The Brain and ThinkMap offer software that automatically displays Web content as an undulating tree of nodes. The interface creates the illusion of three-dimensional spatial perspective, and enterprise content is presented as a giant web. Click on any node and it bounces to the center of the rearranged tree structure. The selected term sits in the center of the spider web and related categories are illustrated with links. The goal is to zero in on a particular section of a hierarchy without losing the context of the surrounding parent/child relationships. Any term within a taxonomy can be highlighted while retaining the rest of the hierarchy as context.

Inxight's Hyperbolic Tree, for example, orders information into a hierarchy, lays out the parent/child structure on a hyperbolic plane and then maps the plane onto a circular display region. Finally, the structure is translated into a two-dimensional space for display on a computer screen. Researchers at Xerox Parc were inspired by a woodcut entitled Circle Limit IV (Heaven and Hell), by the Dutch graphic designer M. C. Escher (see illustration). In this brilliant design, a circle of demons intersects with a circle of angels. The figures diminish in size as they expand outward and the number of figures increases exponentially. Researchers noted the "fisheye" distortion of space in Escher's woodcut and his ability to consistently depict an exponentially growing structure. Hierarchies increase in size exponentially. In essence, Escher's woodcut is a representation of a hyperbolic plane.

Benefits and applications Browsing: When researchers go straight to a document without browsing through nearby material, their labor is efficient but thin (and maybe superficial). The power of research comes from being surprised by things you weren't explicitly looking for. Directories and animated taxonomies encourage casual browsing. For example, while searching for information on JavaScript, you see an interesting link about Computer Programming that is more valuable.

Content Management: Because users access documents by subject or topic, they are able to find the information they need without having to know its location on the company intranet. In addition, a content editor or portal manager can use a topical directory when looking for an appropriate place to catalog company information.

Brainstorming and problem solving: The arrangement of terms within animated taxonomies and hierarchical directories facilitates associative thought and relational learning. They help users reason more clearly. Because these information structures place ideas in context, users ask more focused questions. For example, if I wanted to identify some of the major technologies animating the Internet search market today, a directory could list my options: spiders, information extraction technology, automatic classification software, visualization tools, etc. Hierarchies nudge the user toward more clearly identified research objectives. Aesthetics: Animated taxonomies feature elegant interfaces that are a joy to use. Their spare, modern graphics encourage interaction and use. For example, ThinkMap's animated thesaurus was used by the Smithsonian Institute's first exclusively online exhibition, "Revealing Things" (http://web2.si.edu/revealingthings). The thesaurus operates according to mathematical algorithms derived from water flow patterns. Indeed, ThinkMap claims its product is an artistic spectacle in which words dance in elaborate information choreography.

Improved communication: Directories and taxonomies function to hook people and context together. They provide a common language, and workers better relate concepts across departments, divisions and companies. For example, what one company calls "CRM" another may term "infrastructure management," "one-to-one marketing" or "front-office applications."

Collaboration: Structures of information organization promote collaboration by matching up users with similar interests. Research is more cumulative when, for example, an analyst studying EU banking policies is linked to an employee researching French fiscal policies.

Information structures: the key to portal success There's no such thing as a silver bullet in planning enterprise portals. A visualization or categorization product will never atone for poor information architecture. Portal design requires clear Web site structure, consistent labeling schemes and effective search tools. Even with relevant content and a winning interface, portals will fail unless workers can easily retrieve corporate knowledge.

Portal construction is all about creating structure. They should be thought of in the same way as physical buildings. In both physical and virtual space, information must be organized in a way that makes sense to the people who use it. For example, most people can easily access information in a book because they are familiar with its structure. Because the underlying structure of books (e.g., the table of contents, index and footnotes) is predictable, its organization scheme is invisible to most readers. This familiarity allows the book's content to dominate the reader's attention. In a similar way, directories and animated taxonomies help users make sense of data. Digital information is accessible and useful to the degree that it’s placed in an organizational scheme.

Terms Hyperbolic plane: A hyperbolic plane is one of the most famous examples of non-Euclidian geometry. In Euclidean geometry there is one line passing through a given point, but in hyperbolic geometry there are many. The hyperbolic plane is a mathematical model in which parallel lines diverge away from each other. This means that the circumference of a circle on a hyperbolic plane grows exponentially from its radius. M.C. Escher is famous for creating visual representations of non-Euclidean geometry like the one that illustrates this essay.

Information architects: These professionals combine Web design, library science and technical skills to order enterprise knowledge. Information architects design organizational systems within Web sites that help people find and manage information more successfully.

Katherine C. Adams is an information architect and free-lance writer, e-mail kadams@mohomine.com.


Search KMWorld

Connect