Cooperative Answers in Database Systems
Terry Gaasterland, Parke Godfrey, Jack Minker, and Lev Novik
The complete paper is available in
Abstract
A major concern of researchers who seek to improve human-computer
communication involves how to move beyond literal interpretations of
queries to a level of responsiveness that takes the user's misconceptions,
expectations, desires, and interests into consideration.
At Maryland, we are investigating how to better meet a user's needs within
the framework of the cooperative answering system of Gal and Minker.
We have been exploring how to use semantic information about the database
to formulate coherent and informative answers.
The work has two main thrusts:
- the construction of a logic formula which embodies the content of a
cooperative answer;
and
- the presentation of the logic formula to the user in a natural language
form.
The information that is available in a deductive database system for
building cooperative answers includes integrity constraints,
user constriants, the search tree for answers to a query, and false
presuppositions that are present in the query.
The basic cooperative answering theory of Gal and Minker forms the
foundations of a cooperative answering system that integrates the new
construction and presentation methods.
This paper provides an overview of the cooperative answering strategies
used in the CARMIN cooperative answering system,
an ongoing research effort at Maryland.
Section 2 gives some useful background definitions.
Section 3 describes techniques for collecting cooperative logical
formulae.
Section 4 discusses which natural language generation techniques are useful
for presenting the logic formula in natural language text.
Section 5 presents a diagram of the system.