Natural-language help in the Consul system

  • Authors:

  • William Mark

  • Affiliations:

  • USC/Information Sciences Institute, Marina del Rey, California

  • Venue:
  • AFIPS '82 Proceedings of the June 7-10, 1982, national computer conference
  • Year:
  • 1982

Quantified Score

Hi-index 0.00

Visualization

Abstract

If we use the model of asking an expert, it is fairly clear what users want from a help system: a service that tells them how to do something they want to do. But current help systems aren't like this. They can tell the users about system capabilities, but not in relation to what they want to do. Most help systems are simply databases of online documentation---system manuals, not system experts. Like system manuals, using them to figure out how to do something or to figure out what went wrong is a last resort---when no one else is around. Providing real expert help requires reasoning in terms of models of what the user wants to do and what the system can do. These models also make it possible to provide facilities for natural-language understanding and generation. Thus, users can deal with the system in much the same way that they deal with a human expert: by asking questions and receiving advice in English. These ideas are being tested in the Consul system, a research prototype currently under development at the USC Information Sciences Institute.