Highlighting User Related Advice

Research on explanation techniques for expert systems has demonstrated 由at (1) explanations are most effective when 由ey address the user' s needs and (2) it is necess缸Y to augment explanations with information 由at is missing from the expe口 system's reasoning. It is our thesis 由at explanation content can a1so be improved by removing extraneous information from the system's reasoning and reorganizing 由e remainder to emphasize user concerns. To test our ideas , we have developed an interactive natural language problem-solving system called ADVISOR which advises students on course selection. Previously, we have repo口ed on our methodology for deriving user goa1s from the discourse, representing different points of view in 由e knowledge base and inferring user-oriented advice with a rule-based system 由at employs information from the appropriate perspective to address user goa1s. In this paper, we describe a model for pruning an explanation to highlight the role of the user's goal.叽le model is part of ADVISOR's natural language generation component We demonstrate its efficacy with ex缸nples of different advice that ADVISOR provides for the same query in the context of different goals. WORD COUNT: 3487 (3118 in text, 377 in display). In addition, Figure 1 contains a diagram of approximately 1/2 page. TOPIC: Natural Language