The KM Technology Infrastructure.

Knowledge Management (KM) is a business process and a professional discipline. KM describes a set of business practices and technologies used to assist an organization to obtain maximum advantage from one of its most important assets -- knowledge. Evaluating the technology needed to support this important business process requires a deep understanding of its broad applicability and the potential reach of its scope.

The KM process requires technology to support the capture and sharing of people's knowledge, promote collaboration, and provide unhindered access to an extensive range of information. Technology must support all activities involved in the knowledge life cycle (e.g., capture, organization, retrieval, distribution, and maintenance). To quote Michael Zack (1998), "The information technology infrastructure should provide a seamless pipeline for the flow of explicit knowledge through the five stages of the refining process to enable

* capturing knowledge

* defining, storing, categorizing, indexing, and linking digital objects corresponding to knowledge units

* searching for (pulling) and subscribing to (pushing) relevant content

* presenting content with sufficient flexibility to render it meaningful and applicable across multiple contexts of use.

KM Systems Architecture

There is no out-of-the-box KM solution. Because it is a synthesis of several disciplines and business practices, KM uses an assembly of technologies. The key to success is a well-designed architecture as the foundation.

In the context of information systems, an architecture identifies a system's components and describes how they interact. Developing the architecture establishes high-level understanding of how systems and subsystems interact. According to GartnerGroup (Austin, et al. 1999), there are three layers (or groupings) in the typical enterprise systems architecture model. They are the following:

* Data layer, a unifying abstraction across different types of data with potentially different storage mechanisms (e.g., relational databases, textual data, video, and audio)

* Process layer, which describes the logic that links data with the use people or systems make of it. In the first instance (people as users), presentation is via the user interface; in the second (systems as users), presentation is via program interfaces used to support application integration.

* User interface (UI) layer, which provides access for people to the information assets of the enterprise via logic incorporated in the process layer

It is easier, however, to consider the KM architecture in five layers:

1) user interface

2) knowledge metamodel

3) knowledge repository (or source repositories)

4) knowledge access tools

5) knowledge management enablers

Figure 1 describes the knowledge architecture and shows how the layers are arranged and how they relate to each other (Duffy 1999).

[Figure 1 ILLUSTRATION OMITTED]

The KM systems architecture does not exist in isolation. The technology needed to support KM may already be incorporated into the overall enterprise technology architecture. It is important to understand how technology unique to KM will relate to and integrate with other technologies and to determine what impact, if any, KM will have on the existing architecture. Using the Gartner three-layer description as a guide, the KM architecture layers are integrated as shown in Table 1.

Table 1 - INTEGRATION OF ENTERPRISE AND KM SYSTEMS ARCHITECTURES

Enterprise Systems Architecture   KM Systems Architecture

User Interface                    User Interface
Data Layer                        Knowledge Metamodel and Knowledge
                                    Map
                                  Knowledge Repository/Repositories
                                  Knowledge Access Tools
Process Layer                     Knowledge Management Enablers

User Interface

The user interface (UI) protects the user from technology complexities. It represents the workspace in which the user lives on a day-to-day basis and provides a consistent window into the applications and data used regularly. In many instances, portals similar to those used to access the Internet (e.g., Yahoo!, Lycos, Excite, or Plumtree) represent the user interface layer.

A well-designed UI must be intuitive, responsive, and valuable to the user. Ease of use is critical; without it, the entire KM system's success is threatened. In today's fast-paced business world, there is little time for extensive training (the only way to overcome technical inadequacies) or for reading manuals. If any system intended for day-today use requires more than one hour of the user's time to absorb basic features and become productive, it is probably too complicated. Using the more complex features may require additional learning time, but that can happen on an as-needed basis.

Knowledge Metamodel and Knowledge Map

The knowledge metamodel and the knowledge map are at the heart of the knowledge management system. They are closely linked with the knowledge repository but because of their importance, they are defined separately in the context of the knowledge systems architecture.

The metamodel contains metaknowledge, which is simply "knowledge about the knowledge." It is what is known about each knowledge object in the repository, including:

* who created the knowledge and when

* the knowledge object's format and media

* the knowledge object's purpose

* how the knowledge was used

* concerns about use

* actions and sequences of events surrounding its existence

* circumstances under which the knowledge was developed

* linkages with other knowledge objects

The information in the metamodel provides the knowledge object's context and much of the knowledge content's value. In order to ensure this information's consistency, clarity, and completeness, it is important to structure its capture and storage. Electronic templates and forms completed at the time the knowledge is created are best since it may be difficult to capture after the fact.

Knowledge Map

The knowledge map (K-map) is the navigational system that enables users to find the answers they seek. It is the primary means of representing the entire collection of knowledge objects, regardless of category or location, and helps to identify the links between existing islands of information.

The K-map must provide an easy-to-use method for users to browse internal and external knowledge holdings. It must be structured to coincide with the way users think about and retrieve information; it cannot be designed with emphasis on any requirement other than its use as a retrieval tool. It must also provide pointers to tacit knowledge, because these form a critical element of the overall knowledge management environment.

The K-map may be the front end of an integrated knowledge management product suite. It may be a component in the knowledge portal that serves as the user interface or it can be custom developed. Regardless of how the K-map is implemented, it must provide visual feedback to the user about several (but not too many) hierarchical layers. A model with nine categories at the highest level and a limit of three layers to each successive level will have a total of 36 categories.

The highest levels must represent the organization's business model, using predetermined high-value knowledge categories as the starting point. The ongoing extension of the K-map will evolve as it is used. GartnerGroup suggests (Rosser 1999), "A best practice for optimal creation of the essential knowledge map is to manually build a high-level structure, guided by enterprise usage and consistent rules or principles, and then use that framework to enable the subsequent classification task to be done through automated means."

Knowledge Repository

Knowledge repositories can be fileservers, database servers, groupware servers, document management systems, or Web sites. Each repository's structure will depend on the content or knowledge it stores and manages. Although each is a separate physical repository, together they form a single, virtual knowledge repository.

The taxonomy provided in the knowledge map and the tools housed in the architecture's knowledge access component govern access to the repository's content.

Knowledge Access Tools

The components needed to manage knowledgebase access are a complex combination of system administration tools and knowledge management enablers. The specific tools needed differ, depending on the environment. Some potential categories include:

* security models

* directory interfaces

* network interfaces

* information about knowledge's physical location

* type of database in which knowledge resides

* required protocols

* access tools and engines

* distribution tools and engines

Functionality

The KM technology environment must provide certain basic functions. The level of sophistication needed will depend on KM objectives, and existing technology will provide some desired functions. Technologies that have already proven their value to KM include intranets, groupware, document management, data warehousing, artificial intelligence, and intelligent agents. KM life cycle functionality includes the following functions.

Acquisition and Capture

Knowledge is either created inside an organization, or it is acquired from external sources. As new knowledge is generated or acquired, it is added to the enterprise's knowledge base. Capture technology supports the process of transitioning tacit to explicit knowledge so that it can be shared with others. The technology can mitigate cultural difficulties associated with this transition by providing appropriate language, formats, templates, and models to support the articulation process.

Technology should support the collaborative process and therefore new knowledge generation. Traditional tools for capturing knowledge include word processing, spreadsheets, e-mail, and presentation software. Increasingly, newer technologies such as voice recognition, shared workspaces, and video conferencing will be used to support the knowledge-capture process.

Ideally, the capture technology supports other knowledge life cycle elements, for example, gathering information about the document (e.g., the knowledge container regardless of its form or format) and its creation -- author, date of origination, type of technology used -- as well as assisting in its organization by managing versions, shared authoring, links to related documents, etc. Organizing processes such as indexing, abstracting, integrating, and reclassifying are best applied during (or immediately following) the capture process.

Organization and Storage

The knowledge store or repository acts as a bridge between capture and retrieval processes. Its structure must accommodate multiple content views so that users can examine stored knowledge according to their own contextual needs. According to Zack (1998), "A high degree of viewing flexibility enables users to alter and combine views dynamically and interactively and to more easily apply the knowledge to new contexts and circumstances." Because much of the value in managing knowledge is produced by relating various knowledge elements or objects, the repository structure should support their linking and cross-referencing.

The results of the organizing processes described earlier are stored with the knowledge and provide the context surrounding the knowledge object. The knowledge repository is unlikely to be a single physical entity; in most cases it will consist of several logically linked repositories -- each chosen to suit the particular knowledge or content type stored. For example, records management indexes, customer information, and marketing literature might be stored separately but viewed as though they reside in a single repository.

The classification scheme (represented in the knowledge architecture as the knowledge map) must provide a means for users to effectively navigate the structure. The classification scheme may use hierarchies, taxonomies, or semantic networks, but whatever the method, it must reflect the user's cognitive model of the information and provide an easy-to-understand visual path.

The knowledge base's contents will change and grow over time; some knowledge will be superseded by new knowledge. Documents will need to be removed or archived. As part of the classification scheme, each knowledge element should be assigned a currency indicator and possibly an expiration date. When assigning an expiration date, however, it is important to consider the knowledge's value in the context of its present and potential long-term use. It is a mistake to limit this assessment to the knowledge object's applicability to current or near-future circumstances.

Retrieval

Both tacit and explicit knowledge must be readily available to users through capabilities such as

* easy search and query regardless of content, information source, or location

* efficient access through well-structured taxonomies and indexes

* good quality search tools

* pointers to people with experience, expertise, and knowledge to share

* incorporation of rich media and content (text, video, audio, and images)

One of the KM environment's most important functions is to provide users with a rich knowledge resource, allowing them to find anything that helps find answers. This suggests that users must be able to access knowledge regardless of its domicile; for example, to facilitate both input and retrieval, there should be a direct link from the user interface to desktop and back office applications. Effective, efficient search and retrieval depends on appropriate storage mechanisms, multiple search techniques, well-developed indexing and classification schemes, and transparent access to disparate data sources.

Typically there are two ways to index electronically stored information: full text indexing, which automatically indexes every word, and structured indexing, which is based on prespecified attributes or keywords such as author, title, date, and specific words in the text. Both methods have advantages and disadvantages. A full-text search without any limiting parameters potentially delivers high volumes of irrelevant material. Structured index searches are limited to retrieving material that complies with data entered in the index fields. The optimal answer lies in a combination of both methods, along with taxonomies that reflect the enterprise's business.

Knowledge workers today are required to solve problems and recommend solutions, a far different situation than providing simple responses to simple questions. Therefore, the knowledge management environment must provide sophisticated support for complex work. Ideally, the technology will support multiple search techniques that include:

* natural language searching

* boolean searching 

* automatic root expansion

* thesaurus integration

* search by object type, index fields

* concept searching

* fuzzy searching 

If users can be grouped automatically into work-related or other interest communities or aligned with knowledge objects, then preassembled knowledge collections can be made available to specific user groups based on their interests, and content filters can reduce the volume of irrelevant information provided.

Grouping users by interest community strongly supports new knowledge generation through structured, threaded conversations. Some technologies support alignment of users based on self-selection or peer-selection, with enterprise-determined or automated alignment also possible. In some instances, the system will employ automated agents to learn an individual's preferences and deliver knowledge to them at the point of need.

Distribution and Presentation

Distribution and presentation functionality refer to methods for routing and delivering knowledge to users, often a combination of e-mail, workflow, and "push" technologies. It is useful if the distribution mechanism automatically notifies users of content changes, such as newly posted information and obsolescence of published material.

Intelligent push technology selects what it will automatically send to a user based on "learned" patterns of behavior and information stored in the user profile. There is no limit to the knowledge the agent can search; it is not restricted to internally stored information. The user may be able to specify where the agent should look for the desired knowledge.

New knowledge is the result of interpretation, analysis, and redefinition of context. So far, most of these activities have depended on human intervention. However, there is increasing interest in technologies that support automated discovery of new knowledge through synthesizing search results -- providing users with the results of first-phase analysis in graphical format.

Presentation technologies range from traditional business intelligence products to automated discovery techniques (e.g., data mining, skill mining, text mining). Although automatic visualization of trends and patterns is still far from a mainstream knowledge management capability, technologies that support rich analytical processes will increasingly be part of the knowledge management environment.

Maintenance

There is some debate about whether knowledge diminishes in value as it ages. Theoretically, one could argue that all knowledge is valuable regardless of its age. However, the real world of business needs a means to ensure the knowledge base is kept ever green.

It is important that the system can apply intelligent retention rules to stored knowledge. Some routine purging or archiving work can be delegated to intelligent agents, but human intervention is required to ensure that rules governing the agents' behavior are current.

Without maintenance, the knowledge base is at risk of becoming cumbersome and irrelevant. Organizations are already experiencing this as they implement intranets without sufficient forethought. According to Forrester Research (1999), "Intranets are in chaos ... the majority of Fortune 1000 intranets suffered from unmanaged content ... As new content publishing standards ... make it easier to create HTML documents, employees will be forced to handle ... even more content."

Moving Toward KM Solutions

KM's major objective is to connect people with people and stimulate collaboration. The overall architecture and functionality must support this at all times. The ability to capture and manage human-added value makes information technologies particularly suited to dealing with knowledge.

Knowledge management benefits significantly from real-time communication between individuals that encompasses information exchange and provides a shared workspace (e.g., application sharing and video-conferencing). Such communication accommodates shared creation of work products -- documents, group decision support, networked virtual meetings, etc. -- and provides the ability to link immediately to experts if they are online.

To support both explicit and tacit knowledge, KM solutions need to be built with content and collaboration technologies. Data technologies are structured and typically numerically oriented; knowledge technologies deal most often with text. Knowledge technologies are more likely to need human interaction than data technologies because transactions are fewer and volumes lower, with interaction and iteration commonplace.

There is no single technology that provides all the desired functionality, but it is possible to build a KM infrastructure from multiple technologies. A word of caution: technology is a small piece of the KM puzzle, and it is the wrong place to start.

References

Austin, Tom., A. Cushman, Jackie Fenn, Kathy Harris, and Simon Hayward. "The Impact of Knowledge Management on Enterprise Architecture." Strategic Analysis Report. GartnerGroup Inc. 25 October 1999.

Duffy, Jan. Harvesting Experience: Reaping the Benefits of Knowledge. Prairie Village, KS: ARMA International, 1999.

Forrester Research Inc. Building An Intranet Portal. January 1999.

Rosser, B. Knowledge Mapping -- Automated or Manual? Research Note, GartnerGroup Inc. 4 May 1999.

Zack, Michael H. Managing Codified Knowledge. 1998 Available at www.cba.neu. edu/~mzack/articles/kmarch/kmarch.htm (accessed 22 February 2000).

Jan Duffy is vice president of specialty practices with LGS Group, a management and information technology consulting firm in Canada. She is a senior business improvement professional with a special interest in designing total systems -- people, processes, and technology -- to support knowledge work and knowledge workers. Duffy is a frequently published author on the topic of knowledge management. The author may be contacted at [e-mail].