Impertinent Question-Answering Systems: justification and theory

ABSTRACT

A Question-Answering System (QAS) is impertinent if it tells a user something the user did not ask! Since a user rarely understands the full implications of a data base, especially a large one, his question is limited by his knowledge of the data base. Sometimes, a question similar to the one he originally asked could better satisfy his goals, but he did not ask it because he had no reason to do so. Hence, impertinence is a necessary feature of sophisticated QAS. We describe a complete theory for impertinent QAS. The QAS is impertinent when, and only when, a small change in the question results in a large positive change in the answer, i.e., in the presence of a discontinuity. (The metrics are user dependent.) In impertinent QAS, the user's: “Why didn't you tell me?" answered by “BUT YOU DIDN'T ASK!," no longer occur. The theory has been implemented in a sophisticated airline tariff system.

REFERENCES

• 1: ACM Computing Surveys, March 1966.
• 2: ACM SIGART Newsletter, No. 61, 1977.
• 3: Siklóssy, L., Question-Asking Question-Answering Systems, in Proc. International Seminar on Intelligent Question-Answering and Data Base Systems, J-C Simon and L. Siklóssy, eds. (IRIA, Rocquencourt, France, 1978).
• 4: Kaplan, J., 1978 ACM National Conference Proceedings (this issue).

INDEX TERMS

Primary Classification:
  H. Information Systems
  H.3 INFORMATION STORAGE AND RETRIEVAL
      H.3.4 Systems and Software

Additional Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.7 Natural Language Processing
          Subjects: Language parsing and understanding

General Terms:
Theory

Keywords:
Artificial intelligence, Data base management systems, Discontinuity, Question answering systems, User goals, User model