KE/KA Links

Knowledge Elicitation and Knowledge Acquisition involves acquiring knowledge about how a task is performed. This knowledge consists of both how the task is done and, in some cases, the thought processes used in performing the task. Knowledge Elicitation is a subset of Knowledge Acquisition that specifically refers to obtaining knowledge from a human.

The knowledge acquired has a multitude of uses. One common reason for using KE is to acquire the knowledge used to build an Expert System. Another, more general use, is to obtain data used to write software requirements. This is often done when building a system with a User Interface. It is used less frequently when developing requirements for other portions of the system (which is unfortunate). A third use is during Cognitive Task Analysis where multiple techniques are used to extract information about the thought processes used when performing a task. Engineers who specialize in Knowledge Acquisition are known as Knowledge Engineers.

Techniques for performing knowledge elicitation can be split into two types: direct and indirect. Direct techniques involve asking the subject how they perform the task. Examples include protocol analysis, where the subject is asked to "think aloud" while performing the task, and critical decision method where the user is asked specific questions about why they make certain key decisions. Indirect techniques are used to obtain information that is less easily verbalized by the subject, either because it involves actions performed automatically or because the particular subject has difficulty verbalizing their actions. The most commonly used indirect technique is repertory grid analysis where the subject is asked to list the similarities and differences between entities (such as methods, processes, or results).

Here are some KE/KA related links:

• Methods for Eliciting Knowledge
• Proceedings of the Tenth Kowledge Acquisition for Knowledge-Based Systems Workshop
• Knowledge Acquisition for Knowledge-Based Systems Workshop Archives
• KA/Modelling/Management Mailing List
• KE References
• Protege-II
• EXPECT
• Knowledge Engineering Methods and Languages