PUFF is now in routine use in Presbyterian Hospital, Pacific Medical Center (PMC), in San Francisco. The program produces a report, intended for patient records, that explains the clinical significance of measured quantitative test results and gives a diagnosis of the presence and severity of pulmonary disease in terms of the measured data, referral diagnosis, and patient history. "Rules", or statements of the form "IF condition THEN conclusion ", are used by the physiologist and the computer system to specify the system operation. The sequence of rules used to interpret the case also specifies a line of reasoning about the case, or the detailed explanation of the interpretation of the case. The use of rules for this type of knowledge based system is taken from the results of applied Artificial Intelligence research. In a 144 case prospective evaluation, there was a 91% overall rate of agreement between the rule based system diagnoses and the diagnoses of the designing physiologist; there was a 89% rate of agreement between the system diagnoses and diagnoses of a second independent physiologist.

References 67 October 27, 1977 1 Introduction This application addresses the continuation of research on the applications of artificial intelligence (Al) (1) to experimental molecular genetics. It is an extension of a longstanding effort to cultivate attention to ongoing laboratory research as a domain of explorations in artificial intelligence. Our major effort in this field had been in the DENDRNL project, with analytical organic chemistry as the object discipline.

The performance of a computer-based clinical consultation system is evaluated. The program, called MYCIN, is designed to function as an aid for infectious disease diagnosis are therapy selection, with an initial emphasis on bacteremias. The evaluation methodology is discussed, as well as the difficulties encountered in attempting to evaluate clinical judgments. Specialists in infectious diseases judged MYCIN's final therapy recommendation, and intermediate conclusions about the significance of the infection and identity of infecting organisms. The evaluation techniques described may be useful in assessing the performance of other clinical decision aids. Results of the evaluation show that the program's therapy recommendations meet Stanford experts' standards of acceptable practice 90.9% of the time (Table II), with some variation noted both among individual experts and between Stanford experts and others (Tables I and II).

In collaboration with other members of the MOLGEN project, the author developed a representation system called the "Unit Package" which became operational in July 1977. In some cases (and usually in ignorance), this work has duplicated other representation work that was happening at about the same time. The Unit Package is now being used by several other projects including two away from Stanford. It is written in INTERLISP and runs under the TENEX and TOPS20 operating systems. It is an interactive system for building knowledge-based programs. It also provides a substantial virtual memory so that knowledge bases of several thousand nodes can created without sacrificing the IlTrERLISP environment.

This paper is a "final report" on the first version of the CRYSALIS project. As such, we will summarize the current state of the system and show where we plan to go with it. We have found that a design (in the software eng ineer ing sense) is a valuable tool for the evaluation and augmentation of a program, even when the design is done ex post facto. Using such a design, we discuss the major flaws of the existing system and how to correct them. Finally, we show how the architecture of this system could be useful for certain other task domains.

Counterexamples are provided disproving two independent claims that a simple but accurate method had been found to compute classes of symmetrically equivalent atoms in a molecule. Both methods are good approximations, but are nontheless ad hoc techniques which can sometimes fail to discriminate between atoms which arc, in fact, symmetrically distinct. A recem issue of this journal carried two papers23 dealing loop, at each stage of which a new score is computed for each directly or indirectly with "inexpensive" (in the sense of low atom based upon some function of the current score of that computational effort) methods of perceiving molecular atom arid the current scores of its immediate neighbors. More precisely, each article gives a set of rules term "score" here is used rather loosely; in the case of for scoring2 or comparing3 atoms in a molecule with the claim Morgan's algorithm it is an integer while in that of Shelley that if the scores are equal, or if the comparison shows no and Munk's it is a five-element vector. The important point difference, then the atoms arc symmetrically equivalent (i.e., is that a score is an entity which is associated with an atom can be interchanged by some symmetr) operation on the and which can be compared with the scores of other atoms molecule). Though both methods are doubtless very good in such a way that a strict "greater-less-equal" relationship approximations in the sense that they almost always yield the can be defined.

This is traditionally done with tne aid of a computer programmer acting as intermediary. The dire_t transfer of knowledge from an expert to the system requires a natural-language processor capable of handling a substantial subset of English. The development of such a natural-language processor is a long-term goal of automating knowledge acquisition; faciliting the interface between the expert and the system is a first step toward this goal. This paper describes BAUBAb, a program designed and implemented for hYCIN (Shortliffe 1974), a medical consultation system for infectious disease diagnosis and therapy selection. EAUdAb is concerned with the problem of parsing - recognizing natural language sentences aad encoding tnem into MICIN's internal representation. For this purpose, it uses a semantic grammar in whicft tne non-terminal symools denote semantic categories (e.g., infections and symptoms), or conceptual categories wnicn are common tools of knowledge representation in artificial intelligence (e.g.

An algorithm and its implementation as a computer program is described which for the first time permits the enumeration and construction of all the distinct stereoisomers possible which are consistent with a given empirical formula. The algorithm finds the stereocenters in a chemical structure, takes full account of any symmetry and produces the stereoisomers with cis/trans and R/C designations along with a canonical (unique) name. Examples of its use and a discussion of potential applications are given.

We view this process as occurring in several stages, which together form a data hierarchy (Figure 3.1). The hierarchy offers an overview of DSS function and suggests a task partitioning suitable for a contract net approach.

The term "distributed database" encompasses systems of great diversity. The fact that there is a great variety of distributed databases makes their study and comparison difficult. It is hence important to clearly define what is meant by a given distributed database; it is also important to understand the spectrum of available alternatives. This understanding should elucidate the design of distributed databases and clarify the comparison of different systems. We will define a distributed database in a general way, so as to Include most types of distributed databases: A distributed database is a system that allows integrated access to a collection of logically independent databases [Garcia77].