TruerWords Logo
Google
 
Web www.truerwords.net

Search TruerWords

Welcome
Sign Up  Log On

Knowledge Management, Meta Data, and the Organization

Knowledge Management, Meta Data, and the Organization

My last piece on this subject tried to cover the whole topic of knowledge management in a single page. We'll call that the overview, or the introduction, but it's time to explain a few things in a little more detail.

KM is a hot topic in some circles, but unfortunately it's a phrase that most people misunderstand instinctively. What does the word "knowledge" mean to you? You probably answered with something like "whatever you know", which is true... and that's the problem. When discussing KM, knowledge changes from "something you know" to "something your organization knows".

The goal of a KM system (KMS) is to make the information gathered/produced by each individual (in the organization) available to any and every other part of the organization that might need it, and only when they need it.

For an example we'll again look to our own minds. The human brain is filled with a massive amount of knowledge: the interrelationships of people, places, things, and ideas. All of that knowledge is available to the individual whenever it's needed... or almost so, anyway. Barring damage of some sort, the brain can recall virtually anything, and those things recalled the most often are also recalled quickest.

Just as important as knowledge recall, though, is knowledge blocking, or filters. How could we function if all of our knowledge was "at the front of our minds" all the time? If every thought, experience and task reminded us of everything we know, we'd be paralyzed. That's what the filters do: they keep the unrelated stuff out, and let the relevant knowledge through. (Of course, it's not perfect... the stray thought does zoom through occasionally, but then it's hard to say exactly how "perfect filters" would work.)

Now, back to the organization. All of the different parts of the organization are producing information and developing knowledge, all the time. Think of this organization as an individual brain: the knowledge that's developed by one member or department must be available to other parts of the organization when they need it.

Those who specialize in KM (those who write software specifically for managing an organization's knowledge) would like nothing more than to develop a "hive mind" at the organization. Literally, if such a thing were possible, each member of the organization would be able to share his knowledge with anyone else who needs it, in real time, exactly like a brain cell.

Unfortunately that's not possible, so computers must take the place of the hive mind. When someone in the organization needs to know about something, the computer must be used to translate the request for knowledge from the person's input to something the computer can work with (one or more database queries, usually), gather the information, filter out the irrelevant bits, and present the filtered, contextual knowledge back to the person.

That's the idea, anyway. Try that on the web, and you'll probably be swamped with information. (When's the last time you did a search on the web where most of the results contained what you wanted to find?) Most organizational networks are as useful (useless) as the web, in this regard. There are two reasons: a lack of meta information (information that describes the information), and poorly written searches.

The hypothetical "perfect KM system" will do this as efficiently and effectively as the best human mind. Such a system will probably never exist, but what's amazing is the dramatic improvements that can be made to existing systems with very minor changes. (Even more amazing is how very few organizations take advantage of these things.)

Consider the first of the two reasons that most intranets are so useless: lack of meta information. If an organization has an intranet, then it's probably just full of useless bulletins and other information that's carefully filtered through an editorial department. Those few that actually allow (almost) anyone to post their knowledge have the opposite problem: too much is available, and searching through it involves simple word searches. That's the problem with the web: too much information, and no good way to filter it!

The solution is to add meta data. Even a tiny bit of meta data can make a huge difference! Want proof? Compare Google's search engine with WebCrawler's. Google is much more likely to show relevant links at the beginning of the list because it gives a lot of weight to two pieces of meta data: the page title, and the number of other pages that link to that page.

The page title is meta data. Really! It's not the content of the page, it's a short bit of text that the author has written to describe the contents of that document. So when you search for your name on the web, which page is more likely to contain information about you: a page that's been titled with your name, or a page that mentions your name somewhere in the text? Obviously, the former. Google figured that out better than anyone else, and uses that (along with lots of other things) to make their search engine better.

So to make the information on an intranet useful as knowledge, meta data must be added, indexed, and used. Unfortunately, this is the point where you step out of the theory of KM, and into the hard light of implementation reality: people are lazy.

There are two ways to deal with that fact. First, you can make meta data a required part of the information submission process (remember that in the database it's just info: it's not knowledge until someone uses it) and to do that you must make it very easy to specify meta data. Second, you can let the computer specify the meta data. There's a lot of R&D going into this area right now, everybody with a "high-end" KMS is hoping to use Artificial Intelligence to apply the meta data, freeing the individuals from a task they usually won't (and don't) do if given the choice.

This is a frustrating problem, because meta data is usually easier to write than the data itself. For example, a keywords field on this article would probably contain "organization knowledge management km kms meta data meta data artificial intelligence". That only took about 30 seconds, compared with hours to write this article, and typing it into a special "keywords field" is just as easy as typing it into the article itself. Unfortunately making it easy isn't enough: people need to believe it's important, and know that it's required. Buy-in and enforcement.

What type of meta data is needed? What's the best KMS (or at least the best type of KMS) for the organization? These and other questions wil have to wait for my next article.

Page last updated: 7/4/2002




TruerWords
is Seth Dillingham's
personal web site.
Read'em and weep, baby.