GOLDEN MEANS: KNOWLEDGE MANAGEMENT, DATA MINING AND THE LONG ROAD TO WISDOM
by Inderpal Bhandari, executive editor at large

In past columns, I described Virtual Gold's conference, "The Evolution of Data Mining: Technical Strategies to Beat your Competition by Year 2000," and expanded on some of the messages from the conference. We then took a hiatus of two weeks to discuss other subjects of current interest. Let us now return to a key message from Virtual Gold's data mining conference, specifically, the close relationship between data mining and knowledge management.

Knowledge management is an emerging area. Often times this is simply another way of saying that at the present time we really do not understand the subject. But I think William Woods has a fair understanding. I came across a good description of knowledge management in a recent article written by him in the Sun Journal.

Woods advocates that knowledge management should be viewed as an overarching concept that combines a management philosophy with data warehousing, workflow strategies, database management, and knowledge distribution. He further points out that knowledge management is becoming a concern for CTO's and CIO's, since an organization's collective knowledge -- and the ability to access it -- comprises a key corporate asset.

A down-to-earth way to understand the need for knowledge management is to go back to a familiar saying about a large European multinational company. I shall not name names here. The saying is true for large organizations in general, and it is quite unfair to focus on any one organization. Instead, let us refer to it as E-Co. The saying goes as follows: If E-Co knew what E-Co knows, then E-Co would be a great company. Knowledge management is an attempt to address E-Co's problem.

Data mining has to do with finding unexpected patterns from large amounts of data that eventually lead one to counter-intuitive insights about one's business. It allows one to discover knowledge through means that are largely automated. In a big corporation, several departments could make use of such means.

Consequently, it is not unusual to find large numbers of insights. I have personally been witness to a situation in a large organization where the use of data mining on one major process alone led to 63 findings in the space of six months. Admittedly, not all insights were created equal -- some were far more significant than others. However, they were all useful in that they improved some aspect of the business.

Such knowledge must be preserved, else an organization will spend all its corporate energy reinventing the wheel. Therein lies the link between knowledge management and data mining. If data mining activities are not to become one-off projects, they must be tied to a framework for knowledge management. Else, those hard won insights will become part of what I like to refer to as organizational memory loss -- a corporate liability, much as collective organizational knowledge is a corporate asset, in keeping with William Woods' terminology.

Given that knowledge management is currently more a concept than a solution, tying data mining to knowledge management is easier said than done. There are major challenges that must be addressed. For instance, how does one cope with contradictions between historical knowledge and current knowledge? The data mining patterns stored in the knowledge management framework suggest one action, but the most recent data contradict that suggestion. Such contradictions are a signal that something in the environment has changed, which happens often enough in today's business world. Clearly, if data mining patterns are going to be stored and used over time, we must devise mechanisms to deal with such contradictions.

In summary, I use the words of Frank Zappa: Information is not knowledge, knowledge is not wisdom, etc. Data mining helps us change information to knowledge. Knowledge management will eventually help us keep track off that knowledge. For that reason, it is imperative that those involved in data mining also get involved in knowledge management. But we are still a long way from wisdom.

---
Inderpal Bhandari can be reached via http://www.virtualgold.com