Knowledge Management: How Do You Know What You Know?

Global organizations, government agencies, and millions of users are buried in their own information. It's not just that PC technologies, networks, and the World Wide Web have brought about a proliferation of data. What can be more overwhelming is that it's all typically scattered across a range of repositories--file servers, document management systems, group-ware applications, intranets, relational databases, the Web, and, yes, even paper. Even if each of these islands of information has some search capability, they usually require separate access. This makes finding information on a given subject a daunting task. Fortunately, new knowledge management technologies now maximize the knowledge, workers' efficiency, and effectiveness by enabling them to pinpoint and retrieve information quickly with a single search across a wide range of information sources.

Defining Knowledge Management

Knowledge management (KM) is more than a fad. It's a mandatory practice for any organization. KM can be viewed as a three-legged stool. Take away one leg, and it won't stand up. The three legs are people, processes, and technology. People, because only human beings can have knowledge. Processes, because the transmission of knowledge from person to person requires rules and procedures in order to make use of knowledge or take action on it. Technology is necessary to store, retrieve, and organize vast quantities of information and make it digestible by human beings.

We live in a knowledge economy. To varying degrees all organizations rely on the knowledge of their employees. Take the pharmaceutical industry, for instance. Although they sell drugs and medications, what they're really purveying is the knowledge and creativity of the researchers who develop these products, obtain approval, and get them to market. Typically, only one out of 30 new projects results in marketable medication. The knowledge the researchers carry from project to project therefore has a direct bearing on profits. A company that can improve the ratio to one in 28 through more rapid and efficient development has a competitive advantage.

In the 1970s and '80s, computers were used to transform data--numbers or small bits of textual data--into information by assembling and sorting it. Today's challenge lies with turning information into knowledge and managing that knowledge.

Concept-Based And Pattern-Based Retrieval

One of the two major KM technologies is concept-based retrieval. It determines what a searcher is looking for and extracts the most relevant documents or items. This requires a certain level of sophistication in the search software because individuals use different words to describe what they are looking for. When someone uses the word "bank," for example, the information they're looking for may actually be in a document about Savings and Loans. KM software recognizes such idioms and retrieves information on the entire range of institutions. It also distinguishes between "bank" in a financial sense and all the other meanings of the word. Going even further, the software takes contextual evidence from each document to promote or demote it in importance.

The other essential KM capability is pattern-based retrieval or pattern recognition. When a search item is misspelled or does not match spelling in the source document, the software supplies a list of possible spellings or finds documents that approximate the spelling. Both errors and legitimate variations are covered because the software recognizes similar patterns.

An offshoot of pattern-based technology is used in image retrieval. The searcher supplies an image, such as a rug pattern, and the software finds a similar image. It is also applied to linear assets such as video. A video stream is broken up into scene changes based on pattern changes, essentially cinematic effects like cuts and dissolves. Scenes are then placed in an array the users can browse to find the scene for which they're looking. Cutting edge organizations are now using video indexing and retrieval for KM projects. The role of video as a knowledge asset is growing rapidly, but without pattern recognition, accessing a specific piece of information in analog video is still an inefficient process.

According to AIIM (The Association for Information and Image Management), a typical knowledge worker spends anywhere from 15 to 35 percent of their time looking for information. Because knowledge workers tend to be highly compensated, shortening the time they spend searching for information can have a great impact on the bottom line.

Seamless Searches

Superior knowledge management technology is about time, accuracy, and the ability to share information across an organization. Accessing many separate repositories to locate and compile scattered bits of information wastes time. When, in an effort to save time, searchers look at only a subset of hits, they may miss the most relevant documents because the search engine put them at the bottom of the list. Accurate and complete search results are critical for solving problems and meeting organizations' needs. KM technology spans across different types of repositories, indexes and ranks them according to the interest of the searcher. It makes data accessible in a single step, taking the native security of different systems into account.

No matter how sophisticated the software, human discipline remains a necessary element in KM. Only humans can discard obsolete data, evaluate the reliability of sources, and judge the validity of an author's opinion. Cutting-edge software can take the knowledge worker far in searching out only recent information or documents by particular authors. But there must be human attention to managing the repository.

Best Practice For Implementation

Every organization that deploys a KM systems system must decide on how much to bite off in the first deployment. Some organizations chose to are best deployed their system gradually, department by department. A common mistake is biting off too much According to this philosophy. It's it's best to postpone an organization-wide implementation until a subset of employees have become accustomed to the new processes and management has learned how knowledge workers actually use the system. The University of California at Los Angeles, for example, began by deploying KM software in the Chancellor's Office. Incoming mail is scanned for online viewing and staff can respond to or forward mail electronically. Gone are the days of retrieving mail from stuffed inboxes and forwarding it with hand-written Post-It notes. Deployment in other departments is now underway.

Some other organizations have had success launching an organization-wide KM project on day 1. While more challenging from a technical and planning perspective, this approach leverages the organizational commitment in a single large-scale deployment and bypasses any internal political struggles that may result from favoring one department over another. Both approaches can be successful if given the right planning and internal support.

KM software solutions have two significant advantages for transforming data into information and knowledge. They are much more efficient than any Web search engine and provide access across the entire range of silos of information. All types of repositories can now be unlocked with a single key. This means that organizations need no longer be buried in their information. They can get control by converting it into knowledge and applying it for competitive advantage. Sharing enterprise knowledge assets more efficiently brings the power to meet the organization's needs and those of suppliers and customers.

Mark Myers is Vice President, Product Management director of product marketing at Convera Corp. (San Jose, CA).