Many RLL-1 units are directly used by one or more of the RLL-1 functions listed below. These special ones are enumerated below, following a depth first traversal of the RLL-1 Knowledge Base. Diagram #1 portrays a skeleton of this hierarchy, showing the subset relations joining these various classes.

The Most Important Gain: New Knowledge 18 10 Problems of Knowledge Engineering 19 10.1 The Lack of Adequate and Appropriate Hardware 19 10.2 Lack of Cumulation of Al Methods and Techniques 19 10.3 Shortage of Trained Knowledge Engineers 20 10.4 The Problem of Knowledge Acquisition 21 10.5 The Development Gap 21 11 Acknowledgments 22 1 1 Introduction: Symbolic Computation and Inference This paper will discuss the applied artificial intelligence work that is sometimes called "knowledge engineering". The work is based on computer programs that do symbolic manipulations and symbolic inference, not calculation. The programs I will discuss do essentially no numerical calculation. They discover qualitative lines-of-reasoning leading to solutions to problems stated symbolically.

Abstractions include not only small, simple concepts like hierarchies but also more complex notions like concavity and convexity or particles and waves. A model for an abstraction is essentially an interpretation for the symbols that satisfies the associated axioms. Different task domains can be models of the same abstraction (as biological taxonomy, geological time, and organization charts arc instances of hierarchies): or, said the other way around.

This memo contains two papers that deal with medical computing. The first, written for a book on cybernetics and society, examines the range of medical computing systems, plus some of the logistical and human engineering challenges limiting their utility or acceptance. It addresses five recurring themes that characterize the introduction of medical computing systems: 1) the need for the proposed application, 2) the system users, 3) the logistics of system introduction, 4) the required computational techniques, and 5) the required technological resources. In the context of these topics, suggestions are offered for long-range research and resource policies that are appropriate for assuring the development of practical clinical computing. The second paper, presented at a meeting on artificial intelligence in May 1980, takes a more detailed look at the reasons that medical computing systems have had a limited impact on clinical medicine. When one examines the most common reasons for poor acceptance of such systems, the potential relevance of artificial intelligence techniques becomes evident. The paper proposes design criteria for clinical computing systems and demonstrates their relationship to current research in knowledge engineering. The MYCIN System is used to illustrate the ways in which one research group has responded to the design criteria cited.

One relevant computer science subfield, termed "artificial intelligence" because of its emphasis on symbolic reasoning capabilities Our own is the MYCIN system, a program that received poor clinical acceptance. Despite diverse research assists with the selection of antimicrobial therapy for efforts, and a literature on computer-aided diagnosis that patients with infections.' Knowledge of bacteremia and has numbered at least 1,000 references in the last 20 years, meningitis has been acquired from infectious disease clinical consultation programs have seldom been used other experts and,..ncoded in decision "rules" and tables of than in experimental environments. This knowledge is, in turn, used by a program self-evident. Growth in medical knowledge has far that considers a specific case, interacting with the physician surpassed the ability of the single practitioner to master it requesting advice -and generating a therapeutic all, and the computer's superior information poacessing recommendation. By responding to specific questions capacity thereby offers a natural appeal.

The selection of what to do next is often the hardest part of problem solving. This selection can be structured by di,'Mguishing decisions about the problem from decisions about the problem solving process. When planning decisions are structured in this way, we find that many of the most important decisionc, are about the planning process itself. This exercise tends to expose a variety of decisons, which are usually made implicitly and sub-optimally in planning programs with rigid control structures. This paper develops a layered approach for meta-planning, that is, for planning about planning. It is part of a course of research which seeks to enhance the power of a problem solver by enabling it to reason about its own reasoning processes.

Problem solvers need more than just the facts and logic of a problem domain. To work effectively, they also aced meta-knowledge, that is, knowledge about how to use the facts. For example, hierarchical planning uses meta-knowledge to distinguish between the important considerations and the details of a problem. This knowledge is added to the facts and logic of a domain and used to focus the generation of inferences. It relieves a hierarchical planner from trying to deal with everything at once.

This paper briefly outlines the difference between Branching on the bas of response was the first step coward traditional instructional programs and MAI.

The language designer typically designs that language with one particular application domain in mind: as subsequent types of applications are tned, what had originally been useful features are found to be undesirable limitations, and the language is overhauled or scrapped. One remedy to this bleak cycle might be to construct a representation language whose domain is the field of representational languages itself. One remedy to this bleak cycle might be to construct a representation language whose domain is th field of representation languages itself, a system which could then be tailored to suit many specific applications. Toward this end, we (Professor Douglas Lena and 1) have designed and implemented RLL-1, an object-centered2 Representation Languange Language.3 A representation language language (r11) must explicitly represent the components of representation languages in general and of itself in particular.

Cephalothin and play an educational role by reminding the health professional cephaloridine therapy for bacterial meningitis. Ann Intern Med. of critical factors to consider when prescribing therapy 1975; 82:689-93.