Knowledge Bases

Some time ago I started writing a whitepaper on the concept of knowledge management. It was never finished, for a number of reasons, and eventually the motivation for it disappeared. One of the problems I had with it was that whenever I sat down to write about one aspect the topic started to diverge and I ended up talking about something else. Not so good for a tight coherent whitepaper.

But exactly the sort of thing that appears in blogs. So why not explore some of the thoughts here?

The original incentive was a small application that my brother-in-law put together. The purpose, as far as I could tell, was to extract semantics from a digital artefact and hence more closely correlate items from multiple sources. One of the major intended uses was in targeted search. Of course, I may be a bit wrong on that since the descriptions I got were rarely so clear or concisely expressed.

At the same time our team at work were talking about how to induct new people into the fairly esoteric set of tools that we use. In particular how to make sure that key knowledge from our experts did not get lost as they moved on to other things. The company knowledge base is a useless tool which just collects scattered documents of all sorts and vaguely sorts them into general categories. Besides the team knowledge we had gathered was not generally applicable outside.

So how do you draw something meaningful from several hundred artefacts, of many different types. There were documents specific to a very narrow area and overview diagrams which purported to show the high level. Spreadsheets summarising the last twelve months and 100 slide packs which detailed yesterday’s status updates. All of which was essentially useless because it was not possible to find anything specific without talking to the person who put it there – if they remembered themselves.

So the ideas of what constituted a useful knowledge base were on my mind and I made several starts on covering off the key points that it should cover. The first of which is the critical role that searchability plays!

One other aspect is the actual definition of a knowledge base and why one would be useful. Obviously the purpose would drive the which aspects were most important and hence the design of any specific instance.

The best known and most readily available knowledge base is, of course, the internet itself. Almost anything you need to know is out there somewhere and there are a number of very clever people who have dedicated themselves to helping you find it. However the approach taken is to identify the pages (or pages) which most closely match the request.

What do you do when ALL the documents are at least tangentially related to the search topic, or when the information desired is scattered over a number of different artefacts. This is where a well designed knowledge base could be used.