Question-Asking Question-Answering Systems
Laurent Siklóssy · Texas Digital Library (University of Texas) · 1977
We describe conceptual and initial implementation work carried out to investigate a new class of question-answering systems operating on large, complex data-bases. These systems have the following capabilities: 1) they may suggest to the user new questions, similar to those asked by the user, but that may not have occurred to him since he has limited knowledge of the data-base. 2) they answer questions in complex, non-linear data-bases. Hence, this is an effort to investigate data-bases other than the typical inventory data-bases, and their related set-theoretic cum elementary statistics questions. 3) they make suggestions ta the user about information related to the question, but not explicitly contained in the question. Alternate questions are suggested when the user's question is near a discontinuity in the database. In this case, the "Well, you didn't ask me..." syndrome is avoided. The data-base is organized along several dimensions to make such discontinuities apparent. A discontinuity results if a small change in one dimension of a question could result in a great discontinuity in the answer. Illustrative examples from tax laws, freight transportation, design and in particular air fares are given.