Knowledge management building blocks

Like many new technology trends, knowledge management is often on the tips of peoples' tongues before it has sunk firmly into their minds.

Like many new technology trends, knowledge management is often on the tips of peoples' tongues before it has sunk firmly into their minds. For the sake of clarity and to give the term some teeth, industry experts have settled on four components of a knowledge management system. Searching, indexing, content management and collaboration have emerged as hallmarks of such a system, according to industry experts. They claim that only when all four are in place can an agency boast that it has an authentic knowledge management program running. Incorporating those four into agency systems is a colossal undertaking, they warn. It can take years to approach the pinnacle of knowledge management — having the systems, practices and communities of interest in place so workers can share and aggregate information and expertise regardless of where data resides in agency databases, e-mail repositories or customer relationship management (CRM) systems. During the long quest for such tools and techniques, information technology officials may spend a great deal of time trying to figure out where to start. "The term 'knowledge management' is so nebulous," said Catherine Michaliga, director of Army Knowledge Management, a far-reaching strategy to turn the Army into a network-centric, knowledge-based organization. Although embracing knowledge management can be costly and labor-intensive, it often delivers enough incremental gains to keep agencies interested. "The good news is that you begin building a knowledge management system in pieces," said Prabhakar Raghavan, vice president and chief technology officer of Sunnyvale, Calif.-based Verity Inc. In fact, agencies often build other programs using knowledge management building blocks. The General Services Administration's GSA Advantage procurement Web portal, for example, incorporates sophisticated search tools that are part and parcel of most knowledge management systems. "This is a transaction- oriented site, not a knowledge management or CRM Web site," said Al Iagnemmo, e-business division director of GSA's Federal Supply Service. Even so, to support GSA Advantage, which handles more than 1 million purchases a day, the agency incorporated powerful search and indexing capabilities using Verity's flagship K2 Enterprise software. GSA is now adding decision-support features to generate internal reports on trends, such as the rate at which certain products sell, Iagnemmo said. Today, knowledge management goes beyond searching and reporting. "Search was the killer app in the early days of knowledge management," said Mark Myers, vice president of strategic alliances at Vienna, Va.-based Convera. "Then portals grew up around the need for a place to distribute data throughout an enterprise." Since then, knowledge management has evolved into a broader business objective. "Knowledge management is really a way for a private organization or government agency to make better use of the knowledge they generate day in and day out," Myers said. Getting Started Most search systems provide only rudimentary knowledge management functions, yet that is still where most federal agencies should begin, executives say. "The place for government agencies to start is with search in order for them to begin dealing with their silos of information," said French Caldwell, vice president and research director for Gartner Inc. "This is kind of the place to focus, especially if an agency can't get past those harder-to-do [knowledge management] areas, which tend to be collaboration and the ability to work in cross-functional vertical teams." In fact, several companies — including Verity, Convera and Autonomy Corp. — got their starts as enterprise search companies. Early on, those vendors honed their search and retrieval software to outperform standard Internet search engines. Now they are trying to help large companies and government agencies build more knowledge management capabilities into their search platforms. For instance, Autonomy software can identify patterns of words and match them to common concepts. The San Francisco-based company has built its software on complex matching concepts, such as Bayesian Inference, a mathematical method for determining the relationships between documents and other data. Autonomy's use of adaptive signal processing — a pattern-matching technology — enables it to find the "digital essence" of a document, Then "agents" or applets are created "to seek out similar ideas in Web sites, news feeds, e-mail archives or other documents," according to the company. Meanwhile, Convera's RetrievalWare 8 search software is harnessed to a semantic network that recognizes patterns through a process the company calls "dynamic classification." The method allows users to create their own "personalized knowledge space" by defining the logic that will be used to make connections across agency repositories. "These are not automatic connections," said Convera's Myers. Instead, connections are made through semantic associations. Convera also promotes the software's multilingual capabilities and its ability to hunt through a variety of document formats and video, audio and image files. For instance, the Energy Department uses RetrievalWare to scour its archive of photographs, X-ray images, weapons material standards and other information used by DOE engineers. Search capabilities also form the first level of Verity's K2 Enterprise software. The company's Raghavan emphasizes the importance of "parametric" searches that use metadata to scan structured information such as spreadsheets and free text e-mail to drill deeper for more pertinent matches. "This way, a user searching for a wrench can signal that it is a wrench associated with a B-52, bringing in a category perhaps a strategic air command, for example," Raghavan said. Classification Tools Systems integrators or consultants are vital to the arduous task of creating classification schemes and building information taxonomies, or trees of information. "Taxonomies lay out information in a navigational setting," Raghavan said. Information services vendors such as LexisNexis offer premade taxonomies, while search companies claim to have at least the basic taxonomies for all but the most specialized needs. Still, taxonomy-building typically requires a great deal of manual classification and is now drawing heavily on the library sciences. "It takes a very specialized skill set to do this," the Army's Michaliga said. "You've got to work with a subject-matter expert and have the classification skills." Closely related to indexing or taxonomy-building is the addition of content or document management features into basic search tools, said Gartner's Caldwell, whose company has become deeply involved in knowledge management consulting. "What I see most of our government clients looking at right now is document management, content management and records management in particular. These are strong areas of interest," he said. Caldwell also sees vendors such as Documentum Inc., FileNet Corp., Open Text Corp., Hyperwave AG, iManage Inc., Computer Associates International Inc., SAP AG and PeopleSoft Inc. being pulled into the mix because of the growing need to improve the control of documents and the management of records. Although some vendors claim to include basic content management in their software, experts note that companies are partnering with document management specialists such as Documentum and FileNet. The idea is to forge strong ties with those vendors to strengthen a knowledge management system's interfaces with proprietary databases. Involve the Experts Linking documents and content to people is the linchpin of knowledge management. For example, an enhanced search system could give users the names of others in an agency who had requested similar searches and list the documents those users came up with. Or a knowledge management system might list subject-matter experts in its search results. Verity has included the ability to find and collaborate with experts in its K2 Enterprise software, while other companies are forming partnerships to add such functions. Convera, for instance, has teamed with instant messaging company IMlogic Inc. on occasion. But most government customers are comfortable for now with having agency experts simply listed in directories rather than linked to search results. "Organizations typically like to be in charge of their employees' time," Myers said. "They don't tend to want someone hanging out there answering questions." Others agree that the government as a whole is still a long way from adopting the advanced collaboration techniques that are becoming part of private-sector knowledge management efforts. Still, for agencies interested in making experts available, there are several newer players in the market offering expert location or expertise management systems. They include Kamoon Inc., Tacit Knowledge Systems Inc., AskMe Corp. and Sopheon. But as ever, experts warn that the place to start is not with slick new tools but by getting buy-in from agency leaders. Gartner's Caldwell said, "The agencies that have been most effective, if you look at them, are the handful in which the agency heads and [chief information officers] have shown strong support for [knowledge management] and have stressed from the very top that this is an important activity for the agency."