A strategy for near-term success using knowledge-based systems

Abstract

Knowledge-based System (KBS) technologies have been applied to a variety of knowledge-related tasks with varying degrees of success. Differentiating among classes of knowledge-related tasks, based on the amounts of problem-solving knowledge and case-specific data involved, can provide valuable insight into why this occurs. Based on this comparison, four classes of problems are described. One class, of data-intensive tasks, includes problem types that are difficult or impossible for humans to perform, yet may be solved in a cost-effective manner using currently accessible KBS technology. The characteristic features of problems in this class are given, together with an example of a successfully fielded knowledge-based system that solves a problem from this class.

A methodology for evaluating potential KBS (Knowledge-Based Systems) applications

“…However, there are some types of data-intensive problems where the degree of interest depends more on the type and amount of data involved in the task than on the amount of knowledge required. In such cases, a relatively small amount of knowledge can be both very useful and very profitable (Laufmann 1988…”

“…However, data-intensive tasks are typically those that humans have the greatest difficulty solving effectively. The potential for a successful KBS therefore may be higher for some tasks involving a large quantity of data (Laufmann 1988).…”

“…KBS developed for problems involving relatively small amounts of knowledge applied to large amounts of data may provide an attractive opportunity, for near-term success. Figure 3.1 shows a qualitative comparison of these two task characteristics (Laufmann 1988). The measures for knowledge and data used in Figure 3.1 serve only to promote understanding.…”