Page 140 -

Knowledge Capture and Codifi cation 123

based systems. The goal of such systems is to better organize explicit knowledge and
to store it in corporate memory for long-term retention.
Another widely used tool for explicit knowledge coding is the CommonKADS
methodology ( Schreiber et al. 2000 ; Shadbolt, O ’ Hara, and Crow 1999 ), which is a
knowledge engineering methodology centered on fi ve types of models of an
organization:
1. Task model of the business processes of the organization
2. Agent model of the use of knowledge by executors, both human and artifi cial, to
carry out the various tasks in the organization
3. Knowledge model that explains in detail the knowledge structures and types
required for performing tasks
4. Communication model that models the communicative transactions between
agents
5. Design model that specifi es the architectures and technical requirements needed
to implement a system that embodies the functions detailed by the knowledge and
communication models
In order to implement KADS, the organization is analyzed to identify knowledge-
oriented problems, describe the organizational aspects that may affect knowledge
solutions (e.g., culture, resources), describe the business processes in terms of agents
required, location, knowledge assets deployed, and measures of knowledge intensive-
ness and signifi cance (e.g., mission criticality). Next, the knowledge used in the orga-
nization is described in terms of possessors, processes used in, and whether or not it
is in the right form and location, of the right quality, and available at the right times.
The feasibility of suggested solutions is then checked against the knowledge problems
identifi ed in the fi rst step. This approach allows a systematic cost-benefi t analysis to
be carried out for the processes of knowledge capture.

Decision Trees
Decision trees are another widely used method to codify explicit knowledge. This
representation is both compact and effi cient. The decision tree is typically in the form
of a fl owchart, with alternate paths indicating the impact of different decisions being
made at that juncture point. A decision tree can represent many “ rules ” and when
you execute the logic by following a path down it, you are effectively bypassing rules
that are not relevant to the case at hand. You do not have to look at every rule to see
if it “ fi res, ” and you also take the shortest route to the correct outcome. Their graphi-
cal nature makes them very easy to understand, and they are obviously very well suited