<?xml version="1.0" standalone="yes"?>
<Paper uid="J88-3004">
  <Title>RECOGNIZING AND RESPONDING TO PLAN-ORIENTED MISCONCEPTIONS</Title>
  <Section position="26" start_page="0" end_page="0" type="concl">
    <SectionTitle>
10 CONCLUSIONS
</SectionTitle>
    <Paragraph position="0"> We have presented an explanation-based approach to the problem of recognizing and responding to user misconceptions. The advisor confirms that a user's belief is a misconception by finding an explanation for why he does not hold the user's belief. The advisor infers its source by finding an explanation for the mistaken belief. The process of finding an explanation was presented as one of hypothesizing and trying to verify a small set of potential explanations associated with each type of user belief. In essence, the model uses information about likely sources of different classes of user misconceptions to recognize user mistakes and infer their underlying causes.</Paragraph>
    <Paragraph position="1"> This approach is attractive for several reasons. First, because it has information about classes of abstract misconceptions, it can handle misconceptions of which it has no prior knowledge, as long as they fall into one of these classes. A smaller set of potential explanations can account for a large number of specific user mistakes. Second, the model makes use of knowledge of the types of misconceptions users are likely to make to circumvent the need for general deductive reasoning.</Paragraph>
    <Paragraph position="2"> The model can easily be augmented to first check whether it has knowledge of specific misconceptions, as do tutoring systems, and to use general reasoning when it is confronted with a misconception that cannot be explained by any of its potential explanations, as does SPIRIT. Finally, our approach leads to responses similar (and sometimes more informative) than those of the UNIX advisors we have observed.</Paragraph>
  </Section>
class="xml-element"></Paper>