actions produced in the course cff the explanation. When the user answers questions incorrectly or asks a question of the machine, their apparent level of knowledge decreases, only increasing when they correctly answer questions on a given topic. There are certain problems with the proposed user model: not only is it rather insensitive to the incorporation of new knowledge in the course of an explanation, but it also forces the system to ask the user questions in order to dispel doubts about the user's knowledge when the level of knowledge it has ascertained lies on the boundaries of the expertise levels defined by Cawsey (known, maybe-known mad unknown). Chapter 6 discusses the application of Cawsey's model to a specific problem, where she uses the EDGE system to generate an explanation of how an electronic circuit works. She gives a step-by-step description of each of the system's objects, defining explanation content and structure, each of the system's rules, graphics used for certain explanations, etc., and provides examples of the structure and content of each object and how the system behaves during the explanation process. Finally, in the last chapter of the book Cawsey presents an evaluation of her system,
/lp/association-for-computing-machinery/book-review-neural-networks-for-knowledge-representation-and-inference-cVX0zWF0sV