I spend a reasonable amount of time reading about and looking at Knowledge Management (KM) style software. Lots of different types of programs can be used or abused into performing knowledge management tasks. These range from storing bits and pieces of information in Word Documents or text files (or PDF files!), to Outliners and Notepads with Trees to structure and categorize information, to ever more complex programs that morph trees and display them as graphs, or in other visually exciting and sometimes useful ways. For example programs like Grokster use circles within circles where you drill down deeper and deeper to see things of interest. (Surfulater customers who visit the forums will have seen threads about this there.) The higher end knowledge management tools tend to be quite complex and expensive beasts indeed.

Mind Mapping tools can also be used for KM, but how practical these are I don’t know.

There are several basic problems with KM software which are difficult to resolve. First is the onus placed on the user each and every time new content is added, to somehow categorize the new information. This may be deciding which tree folder it goes in, or what categories or keywords are appropriate for the content or both. This is an obstacle to our productivity, but one that is necessary if we want to readily locate our information in the future.

Our program Surfulater enables users to defer the categorization decision to a later, hopefully more convenient time, so they can add content without anything getting in their way. Deferring the categorization process is just that though, you still need to make time and put the effort in, albeit later on.

One possible technique which can help us out here is automatic classification of text. In essence you read or parse the text trying to pick out specific tidbits of information, which can then used to place it into a pigeon hole.

Another simpler and cruder technique is to apply rules which are used to determine which folder(s) or categorie(s) new content is tied to. A rule can be a simple search string like: if the word engine appears in the text place the information in the ‘car’ folder. Or it can be more complex search terms such as: engine and not boat or truck or tractor. If you’ve ever setup filters in an e-mail program you will be familiar with this technique. These rules can take some time to set up and get working as you desire, but the results are always predictable, unlike other schemes. You can also use rules to enable information to be placed into multiple folders or categories at the same time. We plan to add something along these lines to Surfulater in the future. Time will tell how well it will work in practice, but anything that removes obstacles from our path has to be a plus.

The next difficult area revolves around the ability to easily find information that you know or think you know is in the KM system somewhere. There really isn’t much point saving lots and lots of information if you can’t quickly find stuff again. I’ll leave this discussion for a future post. Stay tuned.