This article examines the direction in which knowledge bases are constructed for diagnosis and decision making. When building an expert system, it is traditional to elicit knowledge from an expert in the direction in which the knowledge is to be applied, namely, from observable evidence toward unobservable hypotheses. However, experts usually find it simpler to reason in the opposite direction-from hypotheses to unobservable evidence-because this direction reflects causal relationships. Therefore, we argue that a knowledge base be constructed following the expert's natural reasoning direction, and then reverse the direction for use. This choice of representation direction facilitates knowledge acquisition in deterministic domains and is essential when a problem involves uncertainty.

Advances in computer hardware and software and engineering methodologies in the 1960s and 1970s led to an increased use of computers by engineers. In design, this use has been limited almost exclusively to algorithmic solutions such as finite-element methods and circuit simulators. However, a number of problems encountered in design are not amenable to purely algorithmic solutions. These problems are often ill structured (the term ill-structured problems is used here to denote problems that do not have a clearly defined algorithmic solution), and an experienced engineer deals with them using judgment and experience. AI techniques, in particular the knowledge-based system (KBS) technology, offer a methodology to solve these ill-structured design problems. In this article, we describe several research projects that utilize KBS techniques for design automation.

Automated knowledge-acquisition systems have focused on embedding a cognitive model of a key knowledge worker in their software that allows the system to acquire a knowledge base by interviewing domain experts just as the knowledge worker would. Two sets of research questions arise: (1) What theories, strategies, and approaches will let the modeling process be facilitated; accelerated; and, possibly, automated? If automated knowledge-acquisition systems reduce the bottleneck associated with acquiring knowledge bases, how can the bottleneck of building the automated knowledge-acquisition system itself be broken? That is, humans are known to be subject to errors and cognitive biases in their judgment processes. How can an automated system critique and influence such biases in a positive fashion, what common patterns exist across applications, and can models of influencing behavior be described and standardized?

This article describes Compaq QUICKSOURCE, an electronic problem-solving and information system for Compaq's line of networked printers. A major goal in designing this system was to empower Compaq's customers with expert system technology, allowing them to solve advanced network printer problems entirely on their own. This process minimizes customer down time; reduces the number of telephone calls to the Compaq Customer-Support Center (resulting in monetary savings); improves customer satisfaction; and, perhaps most importantly, differentiates Compaq printers in the market-place by providing the best and most technologically advanced customer-support facility. This approach also represents a reengineering of Compaq's customer-support strategy and implementation. In its first-generation system, SMART, the objective was to provide expert knowledge to Compaq's help-desk operation to better and more quickly answer customer calls and problems.

In 1991, Ford Motor Company began deployment of CAPE (computer-aided parts estimating system), a highly advanced knowledge-based system designed to generate, evaluate, and cost automotive part manufacturing plans. Many manufacturing processes have been modeled to date, but eventually every significant process in motor vehicle construction will be included. Significant cost reductions are among the many benefits CAPE brings to Ford. CAPE is a highly significant system for Ford of Europe in terms of the business needs it satisfies and the corporate acceptance of AI applications: First, CAPE represents a major investment, with significant person-years of effort spent on predeployment development alone. Second, CAPE is the first large-scale production expert system to be deployed within Ford of Europe.

AI theory and its technology is rarely consulted in attempted resolutions of social problems. Solutions often require that decision-analytic techniques be combined with expert systems. The emerging literature on combined systems is directed at domains where the prediction of human behavior is not required. A foundational shift in AI presuppositions to intelligent agents working in collaboration provides an opportunity to explore efforts to improve the performance of social institutions that depend on accurate prediction of human behavior. Professionals concerned with human outcomes make decisions that are intuitive or analytic or some combination of both.

MetLife processes over 260,000 life insurance applications a year. Underwriting of these applications is labor intensive. Automation is difficult because the applications include many free-form text fields. MetLife's intelligent text analyzer (MITA) uses the information-extraction technique of natural language processing to structure the extensive textual fields on a life insurance application. Knowledge engineering, with the help of underwriters as domain experts, was performed to elicit significant concepts for both medical and occupational textual fields.

We describe an AI system (CTSHIV) that connects the scientific AIDS literature describing specific human immunodeficiency virus (HIV) drug resistances directly to the customized treatment strategy of a specific HIV patient. Rules in the CTSHIV knowledge base encode knowledge about sequence mutations in the HIV genome that have been found to result in drug resistance to the HIV virus. Rules are applied to the actual HIV sequences of the virus strains infecting the specific patient undergoing clinical treatment to infer current drug resistance. A rule-directed search through mutation sequence space identifies nearby drug-resistant mutant strains that might arise. The possible combination drug-treatment regimens currently approved by the U.S. Food and Drug Administration are considered and ranked by their estimated ability to avoid identified current and nearby drug-resistant mutants.

In this project, we have developed the ramp activity coordination expert system (races) to solve aircraft-parking problems. By user-driven modeling for end users and near-optimal knowledge-driven scheduling acquired from human experts, races can produce parking schedules for about 400 daily flights in approximately 20 seconds; human experts normally take 4 to 5 hours to do the same. Scheduling results in the form of Gantt charts produced by races are also accepted by the domain experts. After daily scheduling is completed, the messages for aircraft change, and delay messages are reflected and updated into the schedule according to the knowledge of the domain experts. By analyzing the knowledge model of the domain expert, the reactive scheduling steps are effectively represented as the rules, and the scenarios of the graphic user interfaces are designed.

We are interested in solving real-world planning problems and, to that end, argue for the use of domain knowledge in planning. We believe that the field must develop methods capable of using rich knowledge models to make planning tools useful for complex problems. We discuss the suitability of current planning paradigms for solving these problems. In particular, we compare knowledge rich approaches such as hierarchical task network planning to minimal-knowledge methods such as STRIPS-based planners and disjunctive planners. We argue that the former methods have advantages such as scalability, expressiveness, continuous plan modification during execution, and the ability to interact with humans.