Andrew R. Golding

ICML Conference 1996 Conference Paper

Applying Winnow to Context-Sensitive Spelling Correction

• Andrew R. Golding
• Dan Roth 0001

AIJ Journal 1996 Journal Article

Improving accuracy by combining rule-based and case-based reasoning

• Andrew R. Golding
• Paul S. Rosenbloom

An architecture is presented for combining rule-based and case-based reasoning. The architecture is intended for domains that are understood reasonably well, but still imperfectly. It uses a set of rules, which are taken to be only approximately correct, to obtain a preliminary answer for a given problem; it then draws analogies from cases to handle exceptions to the rules. Having rules together with cases not only increases the architecture's domain coverage, it also allows innovative ways of doing case-based reasoning: the same rules that are used for rule-based reasoning are also used by the case-based component to do case indexing and case adaptation. The architecture was applied to the task of name pronunciation, and, with minimal knowledge engineering, was found to perform almost at the level of the best commercial systems. Moreover, its accuracy was found to exceed what it could have achieved with rules or cases alone, thus demonstrating the accuracy improvement afforded by combining rule-based and case-based reasoning.

AAAI Conference 1991 Conference Paper

Improving Rule-Based Systems Through Case-Based Reasoning

• Andrew R. Golding

A novel architecture is presented for combining rule-based and case-based reasoning. The central idea is to apply the rules to a target problem to get a first approximation to the answer; but if the problem is judged to be compellingly similar to a known exception of the rules in any aspect of its behavior, then that aspect is modelled after the exception rather than the rules. The architecture is implemented for the full-scale task of pronouncing surnames. Preliminary results suggest that the system performs almost as well as the best commercial systems. However, of more interest than the absolute performance of the system is the result that this performance was better than what could have been achieved with the rules alone. This illustrates the capacity of the architecture to improve on the rule-based system it starts with. The results also demonstrate a beneficial interaction in the system, in that improving the rules speeds up the case-based component.