IASKNOT: a simulation-based object-oriented framework for the acquisition of implicit expert knowledge

Research in the field of artificial intelligence (AI) aims to embed aspects of human intelligence in the computer. Several factors constrain the development of a true intelligent autonomous machine. The acquisition of expert knowledge continues to hinder progress. Knowledge acquisition techniques have reduced the effort involved in acquiring knowledge from an expert and representing it in a form that can be used by the computer. Most, however, focus on the gathering and representation of one class of knowledge. Two major categories of expertise makeup most of the expert´s knowledge: explicit knowledge which is easy to articulate, and implicit knowledge such as intuition and judgment. It is in the nature of implicit knowledge that makes it difficult to clearly define and acquire from experts. Most current approaches learn only the expert explicit knowledge via query sessions and ignore the implicit expertise altogether. Humans, on the other hand, continually learn and apply both types of knowledge. Humans typically learn the implicit knowledge by observing others handle real-life situations and by adapting what they observed to handle new situations. This research aims to answer the following question: How does one implement learning by observation such that implicit knowledge can be acquired, represented, and reused? It focuses on capturing and modeling the implicit knowledge that is commonly applied by experts while they deal with dynamic real-life situations. The paper explains the formulated approach and discusses the results obtained from its application to the driving domain