P41 RROO785-09 Appendix C

Appendix c
Community Growth and Project Synopses

This appendix contains a graphical display of the development of the
SUMEX-AIM community over the years and brief synopses of currently active
projects. Figure 15 below illustrates the substantial growth in the
cumulative number of projects in the Stanford, national SUMEX, and Rutgers-
AIM communities since the resource began operation in 1974.

20+ Number of Projects
Rutgers Projects

 

 

 

PON rN
154 ——----seeF?>—+ "7
_ _
10-4 National AIM Projects
a
57-7
Stanford Projects
ETT TTT TTT TT
Jan Jan Jan Jan Jan Jan Jan Jan
1975 1976 1977 1978 1979 1980 1981 1982

Figure 15. SUMEX-AIM Growth by Community

277 E. A. Feigenbaum
Appendix C P41 RROO785-09

National AIM Project: ACQUISITION OF COGNITIVE PROCEDURES (ACT)

Principal Investigator: Jonn R. Anderson, Ph.D.
Department of Psychology
Carnegie-Mellon University
Pittsburgh, Pennsylvania 15213
(412) 578-2788 (ANDERSON@SUMEX-AIM)

The ACT Project combines a semantic network data-base with a
production system to simulate human cognition. Prominent among the reasons
for using a production system architecture as a framework for developing
such a program is the possibility of modeling learning as the acquisition
of new productions. ACT possesses a number of learning mechanisms which
have been used to model the learning of procedural skills such as language
comprehension and geometry theorem proving. Some of these mechanisms have
the effect of either extending or restricting the set of circumstances in
which a particular behavior is performed so as to produce better
performance. Others have the effect of speeding up cognitive operations by
compressing the effects of a series of production applications into the
application of a single production. Out of this set of productions ACT
applies those that usually result in desirable outcomes. In this way it is
able to model the human ability to learn even when given unreliable
feedback. Another feature of ACT that reflects its psychological
orientation is its willingness to model human limitations. Here the hope
is that by being faithful to the human mind even in its failings, it
eventually may be possible to emulate its successes.

SOFTWARE AVAILABLE ON SUMEX

The ACT production system is available to GUEST users of SUMEX.

REFERENCES

Anderson, J.R.: Language, Memory, and Thought. Lawrence Erlbaum Associates,
Hillsdale, N.J., 1976.

Anderson, J.R., Kline, P.J. and Lewis, C.H.: A production system model of
language processing. IN M.A. Just and P.A. Carpenter (eds.), Cognitive
Processes in Comprehension. Lawrence Erlbaum Associates, Hillsdale,
N.J., 1977.

Anderson, J.R. and Kline, P.J.: A learning system and its psychological
~ implications. Proc. Sixth IJCAI, Tokyo, August, 1979.

E. A. Feigenbaum 278
P41 RROO785-09 Appendix C

National AIM Project: CADUCEUS (formerly INTERNIST)

Principal Investigators: Jack D. Myers, M.D. (MYERS@SUMEX-AIM)
Harry E. Pople, Ph.D. (POPLE@SUMEX-AIM)
University of Pittsburgh
Pittsburgh, Pennsylvania 15261
Dr. Pople: (412) 624-3490

The major goal of the CADUCEUS Project is to produce a reliable and
adequately complete diagnostic consultative program in the field of
internal medicine. Although this program is intended primarily to aid
skilled internists in complicated medical problems, the program may have
spin-off as a diagnostic and triage aid to physicians’ assistants, rural
health clinics, military medicine and space travel. In the design of
CADUCEUS and its predecessor INTERNIST I, we have attempted to model the
creative, problem-formulation aspect of the clinical reasoning process.
The program employs a novel heuristic procedure that composes differential
diagnoses, dynamically, on the basis of clinical evidence. During the
course of a CADUCEUS or INTERNIST I consultation, it is not uncommon for a
number of such conjectured problem foci to be proposed and investigated,
with occasional major shifts taking place in the program's
conceptualization of the task at hand.

SOFTWARE AVAILABLE ON SUMEX

Versions of INTERNIST are available for experimental use, but the
project continues to be oriented primarily towards research and
development; hence, a stable production version of the system is not yet
available for general use.

REFERENCES

Pople, H.E., Myers, J.D. and Miller, R.A.: The DIALOG model of diagnostic
logic and its use in internal medicine. Proc. Fourth IJCAI, Tbilisi,
USSR, September, 1975.

Pople, H.E.: The formation of composite hypotheses in diagnostic problem

solving: An exercise in synthetic reasoning. Proc. Fifth IJCAI,
Boston, August, 1977.

279 E. A. Feigenbaum
Appendix C P4i RROO785-09

National AIM Project: HIERARCHICAL MODELS OF HUMAN COGNITION

Principal Investigators: Walter Kintsch, Ph.D. (KINTSCH@SUMEX-AIM)
Peter G. Polson, Ph.D. (POLSON@SUMEX-AIM)
Computer Laboratory for Instruction
in Psychological Research (CLIPR)
Department of Psychology
University of Colorado
Boulder, Colorado 80302
(303) 492-6991

Contact: Dr. James Miller (JMILLERGSUMEX-AIM)

The CLIPR Project is concerned with the modeling of complex
psychological processes. It is comprised of two research groups. The
prose comprehension group has completed a project that carries out the
microstructure text analysis described by Miller and Kintsch (1980),
yielding predictions of the recall and readability of that text by human
subjects. More recently, this group has been interacting with the
Heuristic Programming Project at Stanford, using the AGE and UNITS packages
to build a more complex model of the knowledge-based processes
characteristic of prose comprehension. The planning group is working
toward a model of the planning processes used by expert computer software
designers. The initial development of this model requires the detailed
analysis of expert software design protocols for subsequent simulation.

SOFTWARE AVAILABLE ON SUMEX

A set of programs has been developed to perform the microstructure
text analysis described in Kintsch and van Dijk (Psychological Review,
1978) and Miller and Kintsch (1980). The program accepts a
propositionalized text as input, and produces indices that can be used to
estimate the text’s recall and readability. A more complex model based in
AGE and UNITS, which emphasizes the knowledge-based aspects of
comprehension, is currently under development.

REFERENCES

Atwood, M.E., Polson, P.G., Jeffries, R. and Ramsey, H.R.: Planning as a
process of synthesis. Technical Report SAI-78-144~DEN. Science
Applications, Inc., Denver, Colorado, December, 1978.

Jeffries, R., Turner, A.A., Polson, P.G., and Atwood, M.A.: The processes
involved in designing software. In J.R. Anderson (Ed.), Cognitive
skills and their acquisition. Hillsdale, NJ: Erlbaum, 1981
(forthcoming) .

Kintsch, W.: On modeling comprehension. Educ. Psychologist, 14:3-14, 1979.

Miller, J.R. and Kintsch, W.: Readability and recall of short prose

E. A. Feigenbaum 280
P4i RROO785-09 Appendix C

passages: A theoretical analysis. J. Experimental Psychology: Human
Learning and Memory, 1980. (In press)

281 E. A. Feigenbaum
Appendix C P41 RROO785-09

National AIM Project: BIOMEDICAL KNOWLEDGE ENGINEERING
IN CLINICAL MEDICINE (PUFF-VM)

Principal Investigators: John J. Osborn, M.D.
The Institutes of Medical Sciences
Pacific Medical Center
San Francisco, California 94115
(415) 567-0900 (OSBORN@SUMEX-AIM)

Edward H. Shortliffe, M.D., Ph.D.
Department of Medicine
Stanford University

The PUFF-VM Project has produced two knowledge-based programs for the
interpretation of physiologic measurements made in clinical medicine. The
interpretations are intended to aid in diagnostic decision-making and in
selecting therapeutic actions. The programs are: PUFF--the evaluation of
pulmonary function laboratory data, and VM--the evaluation and management
of respiratory status for patients in the intensive care unit.

The task of the PUFF PROGRAM is to interpret standard measures of
pulmonary function. In the laboratory at the Pacific Medical Center (PMC),
about 50 parameters are calculated from measurement of lung volumes, flow
rates, and diffusion capacity. In addition to these measurements, patient
history and referral diagnosis also are used to interpret the test results.
PUFF produces a report for the patient record, explaining the clinical
Significance of measured test results. It also provides a diagnosis of the
presence and severity of pulmonary disease. The interpretation process is
accomplished by examination of expert knowledge represented by a set of
production rules. Each rule relates physiologic measurements or states to
a conclusion about the physiologic significance of the measurement or
state. A version of the PUFF program is used daily at the PMC.

The VENTILATOR MANAGER (VM) PROGRAM is designed to interpret on-line
physiologic data in the intensive care unit (ICU). These data are used to
manage post-surgical patients receiving mechanical assistance in breathing.
VM is an extension of a physiologic monitoring system, and is designed to
perform 5 specialized tasks in the ICU: 1) to detect possible measurement
errors; 2) to recognize untoward events in the patient/machine system and
suggest corrective action; 3) to summarize the patient's physiologic
status; 4) to suggest adjustments to therapy based on the patient’s status
over time, and long-term therapeutic goals; and 5) to maintain a set of
patient-specific expectations and goals for future evaluation by the
program. The program produces interpretations of the physiologic
Measurements over time, using a model of the therapeutic procedures in the
IcU and clinical knowledge about the diagnostic implications of the data.

SOFTWARE AVAILABLE ON SUMEX
The PUFF and VM programs will be available to GUEST users for use on

pre-existing (non-identifiable) cases. No packages currently exist for
program development.

E. A. Feigenbaum 282
P41 RROO78S~-09 Appendix C

REFERENCES

Fagan, L.M., Kunz, J.C., Feigenbaum, E.A. and Osborn, J.J.: A symbolic
processing approach to measurement interpretation in the intensive
care unit. Proc. Third Annual Symposium Computer Applications in
Medical Care, Silver Spring, Maryland, October, 1979, pp. 30-33.

Fagan, L.M., Shortliffe, E.H. and Buchanan, B.G.: Computer-based medical
decision making: From MYCIN to VM. Automedica 3(2), 1980.

Kunz, J.C., Fallat, R.J., McClung, D.H., et al: A physiological rule based
system for interpreting pulmonary function test results. Heuristic
Programming Project Report HPP-78-164, Computer Science Dept.,
Stanford Univ., November, 1978.

Osborn, J.J., Fagan, L.M., Fallat, R.J., et al: Managing the data from

respiratory measurements. Med. Instrumentation, November-December,
1979.

283 E. A. Feigenbaum
Appendix C P41 RROO785-09

Rutgers AIM Project: RUTGERS RESEARCH RESOURCE -
COMPUTERS IN BIOMEDICINE

Principal Investigator: Saul Amarel, Ph.D.
Department of Computer Science
Rutgers University
New Brunswick, New Jersey 08903
(201) 932-3546 (AMARELGRUTGERS)

The broad objective of the Resource is to apply advanced methods in
computer science, particularly in artificial intelligence (AI), to
biomedical problems. The Resource has three major areas of study: 1)
Medical Modeling and Decision Making in several medical domains with
emphasis on collaborative development of consultation systems in
rheumatology and ophthalmology; 2) Modeling of Belief Systems and
Commonsense Reasoning with emphasis on the psychology of plan recognition
and handling of stereotypes; and 3) Artificial Intelligence studies with
emphasis on Representations, Interpretation processes, and problems of
knowledge and expertise acquisition. The studies in Medical Modeling and
Decision Making are performed jointly by computer and medical scientists at
Rutgers and elsewhere in the Country and abroad.

The Resource also sponsors national Artificial Intelligence in
Medicine (AIM) Workshops for the AIM community.
SOFTWARE AVAILABLE ON SUMEX

CASNET--System for consultation in the diagnosis and treatment of
glaucoma.

Documentation available:

"A Model-Based Method for Computer-Aided Medical Decision
Making" Weiss, S., Kulikowski, C., Amarel, S., Safir, A.,
(1978) pp. 145-172, AI Journal, North Holland Press, 1978.

EXPERT--System for designing and applying consultation models using a
Telatively simple language to describe the models. It has been
used in a variety of medical and non-medical applications
(mainly rheumatology, ophthalmology and endocrinology).

Documentation available:

“A Guide to the Use of the EXPERT Consultation System" Weiss,
S., Kern, K., and Kulikowski, C., CBM-TR-94, 11/78.

"EXPERT: A System for Developing Consultation Models,"
Proceedings IJCAI, Tokyo, August 1979, pp. 942-947, (also
Rutgers Computer Science Report CBM-TR-97) .
Anyone interested in using CASNET or EXPERT may contact either
Kulikowski@Rutgers or Weiss@Rutgers.

E. A. Feigenbaum 284
P41 RROO785-09 Appendix C

REFERENCES

Amarel, S.: Computer-based interpretation and modeling in medicine and
psychology: The Rutgers Research Resource. IN Siler and Lindberg
(eds.), Computers in Life Science Research. Foseb and Planum, 1975.

Schmidt, C.F., Sridharan, N.S., and Goodson, J.L.: The plan recognition
problem: An intersection of psychology and artificial intelligence.
AI Journal 11(1,2), August, 1978 (special issue on applications to the
sciences and medicine).

Weiss, S., Kulikowski, C.A., Amarel, S. and Safir, A.: A model-based method
for computer-aided medical decision-making. AI Journal 11(1,2),
August, 1978 (special issue on applications to the sciences and
medicine) .

285 E. A. Feigenbaum
Appendix C P41 RROO785-09

National AIM Project: SIMULATION AND EVALUATION
OF CHEMICAL SYNTHESIS (SECS)

Principal Investigator: W. Todd Wipke, Ph.D.
Department of Chemistry
University of California at Santa Cruz
Santa Cruz, California 95064
(408) 429-2397 (WIPKE@SUMEX-AIM)

The SECS Project aims at developing practical computer programs to
assist investigators in designing syntheses of complex organic molecules of
biological interest. Key features of this research include the use of
computer graphics to allow chemist and computer to work efficiently as a
team, the development of knowledge bases of chemical reactions, and the
formation of plans to reduce the search for solutions. SECS is heing used
by the pharmaceutical industry for designing syntheses of drugs.

A spin-off project, XENO, is aimed at predicting the plausible
metabolites of foreign compounds for carcinogenicity studies. First, the
metabolism is simulated; then the metabolites are evaluated for possible
carcinogenicity.

SOFTWARE AVAILABLE ON SUMEX

SECS-- An organic synthesis design program available with a reaction
library of over 500 reactions. The program is accessible to users
via a teletype or DEC GT40 type graphics terminal.

XENO-- A program for prediction of metabolites of xenobiotic compounds.
Although the project is still in the early development stages, this
program is available for preliminary exploration and testing.

PRXBLD--A facility for building approximate 3-dimensional molecular models
from their 2-dimensional representations. The program employs an
energy minimization approach and is available both stand-alone and
as part of SECS.

REFERENCES

Spann, M.L., Chu, K.C., Wipke, W.T. and Ouchi, G.: Use of computerized
methods to predict metabolic pathways and metabolites. J. Env.
Pathology and Toxicology 2:123, 1978.

Wipke, W.T., Smith, G., Choplin, F. and Sieber, W.: SECS--Simulation and
Evaluation of Chemical Synthesis: Strategy and planning. IN Computer-
assisted Organic Synthesis Planning. ACS Symposium Series, 1977, pp.
97~127.

Wipke, W.T., Ouchi, G. and Krishnan, S.: Simulation and Evaluation of

Chemical Synthesis--SECS. An Application of Artificial Intelligence
Techniques. Artificial Intelligence 10:999, 1978.

E. A. Feigenbaum 286
P41 RROO785-09 Appendix C

National AIM Project: PROBLEM SOLVING EXPERTISE

Principal Investigators: Paul E. Johnson, Ph.D.
Center for Research in Human Learning
205 Elliott Hall
University of Minnesota
Minneapolis, Minnesota 55455
(612) 373-5302 (PJOHNSON@SUMEX-AIM)

William B. Thompson, Ph.D.
Department of Computer Science

136 Lind Hall

University of Minnesota
Minneapolis, Minnesota 55455

(612) 373-0132 (THOMPSON@SUMEX~AIM)

The Minnesota SOLVER project focuses upon the development of
strategies for discovering and representing the knowledge and skill of
expert problem solvers. Although in the last 15 years great progress has
been made in synthesizing the expertise required for solving complex
problems, most expert systems embody only the limited amount of expertise
that individuals are able to report in a particular constrained language
(e.g. production rules). What is still lacking is a theoretical framework
capable of reducing dependence upon the expert’s intuition or on the near
exhaustive testing of possible organizations. Our methodology consists of:
(1) extensive use of verbal thinking aloud protocols as a source of
information from which to make verbal inferences about underlying cognitive
structure and process; (2) development of computer models as a means of
testing the adequacy of inferences derived from protocol studies; (3)
testing and refinement of the cognitive models based upon the study of
human and model performance in experimental settings. Currently, we are
investigating problem-solving expertise in domains of medicine, science,
and law.

SOFTWARE AVAILABLE ON SUMEX
A version of the DIAGNOSER simulation model is being implemented on
SUMEX. It should be available this summer.

REFERENCES

Johnson, P.E.: What kind of expert should a system be? The Journal of
Medicine and Philosophy. (in press)

Johnson, P.E., Duran, A., Hassebrock, F., Moller, J., Prietula, M.,
Feltovich, P., and Swanson, D.: Expertise and error in diagnostic
Teasoning. Cognitive Science 5:235-283, 1981.

Johnson, P.E., Hassebrock, F., Duran, A., and Moller, J.H.: Multimethod

study of clinical judgement. Organizational Behavior and Human
Performance. (in press)

287 E. A. Feigenbaum
Appendix C P41 RROO785-09

Johnson, P.E.: Cognitive Models of Medical Problem Solvers. IN D.C.
Connelly, E. Benson, and D. Burke (eds.), Clinical Decision Making and
Laboratory Use. University of Minnesota Press. (in press)

Johnson, P.E., and Thompson, W.B.: Strolling down the garden path:

Detection and recovery from error in expert problem solving. Proc.
Seventh IJCAI, Vancounver, B.C., August, 1981, pp. 214-217.

E. A. Feigenbaum 288
P41 RROO785-09 Appendix ¢

Stanford Project: GENERALIZATION OF AI TOOLS (AGE)

Principal Investigators: H. Penny Nii (NII@SUMEX-AIM)
Edward A. Feigenbaum, Ph.D.
Department of Computer Science
Stanford University
Stanford, California 94305
H.P. Nii: (415) 497-2739

The long-range objective of AGE, a SUMEX CORE RESEARCH Project, is to
build a software laboratory for building knowledge-based, application
programs. It is an attempt to define and accumulate knowledge-engineering
tools, with rules to guide in the use of these tools. The design and
implementation of the AGE program will be based primarily on the
experiences gained in building knowledge-based programs by the Stanford
Heuristic Programming Project in the last decade (The programs that have
been or are being built are: DENDRAL, META-DENDRAL, MYCIN, HASP, AM,
MOLGEN, GUIDON, CRYSALIS, PUFF, VM and SACON.). The initial AGE program
contains a collection of tools suitable for constructing user programs
based on the Blackboard paradigm (used in HASP and CRYSALIS) and Backward-
chained production rules (used in MYCIN). In addition, AGE has facilities
to aid the user in the construction, debugging, and running of his program.

SOFTWARE AVAILABLE ON SUMEX

AGE-1 is available on an experimental basis to a limited number of
users. A public version of the programs, together with reference manuals
and user guides, is planned for July, 1980.

REFERENCES

Nii, H.P. and Feigenbaum, E.A.: Rule-based understanding of signals. IN
D.A. Waterman and F. Hayes-Roth (eds.), Pattern-directed Inference
Systems. Academic Press, 1978, pp. 483-501.

Nii, H.P. and Aiello, N.: AGE: A knowledge-based program for building
knowledge-based programs. Proc. Sixth IJCAI, Tokyo, August, 1979, pp.
645-655.

Aiello, N., Bock, C., Nii, H.P. and White, W.: Joy of AGE-ing: An

Introduction to the Use of AGE-1 System, HPP Technical Memo, July,
1980.

289 E. A. Feigenbaum
Appendix C P4t RROO785-09

Stanford Project: HANDBOOK OF ARTIFICIAL INTELLIGENCE

Principal Investigator: Edward A. Feigenbaum, Ph.D.
Department of Computer Science
Stanford University
Stanford, California 94305
Contact: Avron Barr (BARR@SUMEX-AIM)
(415) 497-4719

The AI Handbook Project is a part of SUMEX CORE RESEARCH aimed at
making the important results of AI research accessible to the large, multi-
disciplinary community of scientists who want to build AI systems in their
own problem areas. Students and researchers at Stanford and other AI
laboratories have prepared over 300 short articles describing the
fundamental ideas, useful techniques, and exemplary programs developed in
the field over the last 20 years. These articles have been written for
computer-literate scientists and engineers in other fields who are
unfamiliar with AI research and jargon. The Handbook will provide a
scientist who, for instance, might want to know what a "heuristic" is or
how to build a "natural language" front end, with information about all of
the relevant AI techniques and existing systems, as well as abundant
pointers into the field’s literature.

The Handbook is being published in report and book form. It also
will be made available to the SUMEX community via an on-line information
retrieval system. Following is a TOPIC OUTLINE for Volumes I and II:
HANDBOOK OF ARTIFICIAL INTELLIGENCE

INTRODUCTION: The Handbook of Artificial Intelligence; Overview of AI
Research; History of AI; An Introduction to the AI Literature

SEARCH: Overview; Problem Representation; Search Methods for State
Spaces, AND/OR Graphs, and Game Trees; Six Important Search
Programs

REPRESENTATION OF KNOWLEDGE: Issues and Problems in Representation
Theory; Survey of Representation Techniques; Seven Important
Representation Schemes;

AI PROGRAMMING LANGUAGES: Historical Overview of AI Programming
Languages; Comparison of Data Structures and Control Mechanisms
in AI Languages; LISP

NATURAL LANGUAGE UNDERSTANDING: Overview - History and Issues;
Grammars; Parsing Techniques; Text Generation Systems; Machine
Translation; The Early NL Systems; Six Important Natural Language
Processing Systems

SPEECH UNDERSTANDING SYSTEMS: Overview - History and Design Issues;
Seven Major Speech Understanding Projects

E. A. Feigenbaum 290
P41 RROO785-09 Appendix C

APPLICATIONS-ORIENTED AI RESEARCH~-SCIENCE AND MATHEMATICS: Overview;
TEIRESIAS ~ Issues in Expert Systems Design; Research on AI
Applications in Mathematics (MACSYMA and AM); Research on Al
Applications in Chemistry (DENDRAL, CRYSALIS, etc.); Other
Scientific Applications Research

APPLICATIONS-ORIENTED AI RESEARCH--MEDICINE: Overview of Medical
Applications Research; Six Important Medical Systems

APPLICATIONS-ORIENTED AI RESEARCH--EDUCATION: Historical Overview of
AI Research in Educational Applications; Issues and Components of
Intelligent CAI Systems; Seven Important ICAI Systems

AUTOMATIC PROGRAMMING: Overview; Techniques for Program Specification;
Approaches to AP; Eight Important AP Systems

The following sections of the Handbook are still in preparation and
will appear in Volume III: Theorem Proving; Vision; Robotics; Information
Processing Psychology; Learning and Inductive Inference; Planning and
Related Problem-solving Techniques.

291 E. A. Feigenbaum
Appendix C P41 RROO785-09

Stanford Project: DENDRAL-~RESOURCE RELATED RESEARCH -
COMPUTERS IN CHEMISTRY

Principal Investigator: Carl Djerassi, Ph.D.
Department of Chemistry
Stanford University
Stanford, California 94305

Contact: Dr. James Nourse (NOURSE@GSUMEX-AIM)
(415) 497-1712

The DENDRAL Project involves research in computer-assisted structure
elucidation of unknown organic compounds of biological importance. This
research has three major components: 1) program development; 2) biochemical
applications; and 3) resource-sharing.

Recent program developments have been directed toward building more
powerful interactive programs to assist chemists in the three major areas
of structure elucidation: analysis of data to yield substructural
information about an unknown ("planning"), advanced methods for assembly of
substructures into complete structures ("structure generation"), and the
prediction of data for structural candidates to rank-order the candidates
by comparison of predicted and observed data ("testing"). Important
problems of structure representation have been solved which have enabled
dealing with stereochemical (three-dimensional) aspects of structure
throughout the procedures.

Major areas of application of the programs in the research of this
group and other collaborative projects include: a) marine natural products,
particularly marine steroids and halogenated compounds which display
biological activity; b) antibiotics and other derivatives of known or
potential drugs; c) terpene alkaloids; d) photoproducts related to vitamin
A; and e) conformational studies of narcotic analogs and polypeptides.

These programs are shared among a community of collaborators and
guest users at SUMEX, with communication via computer network from a
variety of sites in the U.S., Europe and Australia. Exportable versions of
some programs are maintained. These versions have been installed
successfully in more than 15 research laboratories throughout the world.

SOFTWARE AVAILABLE ON SUMEX

CONGEN--An interactive program for structure generation to yield candidate
structures for an unknown based on inferred substructural components
(exportable).

GENOA~-An advanced structure generator capable of handling overlapping
substructural information; uses CONGEN as a core component

(exportable) .

Meta-DENDRAL--An INTSUM, RULEGEN and RULEMOD sequence for automatic rule

E. A. Feigenbaum 292
P41 RROO785-09 Appendix Cc

formation to relate observed data to substructures in mass
spectrometry and carbon magnetic resonance spectroscopy.

REACT--A program for carrying out a complex sequence of chemical reactions
and exploration of the consequences of those reactions.

NMR--For substructural inference and spectrum prediction in carbon magnetic
resonance spectroscopy (will be exportable).

REFERENCES

Carhart, R.E., Varkony, T.H. and Smith, D.H.: Computer assistance for the
structural chemist, IN D.H. Smith (ed.), Computer-Assisted Structure
Elucidation. American Chemical Society, Washington, D.C., 1977, p.
126.

Djerassi, C., Smith, D.H. and Varkony, T.: A novel role of computers in the
natural products field. Naturwiss. 66:9, 1979.

Nourse, J.G., Carhart, R.E., Smith, D.H. and Djerassi, C.: Exhaustive

generation of stereoisomers for structure elucidation. J. Am. Chem.
Soc. 101:1216, 1979.

293 E. A. Feigenbaum
Appendix C P41 RROO785-09

Stanford Project: EXPEX -- THE EXPERT EXPLANATION PROJECT

Principal Investigator: Edward H. Shortliffe, M.D., Ph.D.
Departments of Medicine and Computer Science
Stanford University Medical Center - Room TC117
Stanford, California 94305
(415) 497-6979 (SHORTLIFFE@SUMEX-AIM)

EXPEX is a recent Stanford research project that is involved with the
development of new representation schemes to facilitate knowledge
acquisition and explanation. This includes not only the study of
fundamental representational formalisms but also the encoding of various
types of knowledge, such as causal information and user models. The
explanation portion of the research effort is dealing with a medical domain
(endocrinology) and is being undertaken on SUMEX, whereas the knowledge
acquisition portion deals with nonmedical topics and uses other computing
resources at Stanford.

Our interest in explanation derives from the insights we gained in
developing explanatory capabilities for the MYCIN system. In that system
and its descendants, we were able to generate intelligible explanations by
taking advantage of a rule-based representation scheme. The limitations of
the justifications generated using MYCIN’s explanation capabilities have
become increasingly obvious, however, and have led to improved
characterization of the kinds of explanation capabilities that must be
developed if clinical consultation systems are to be accepted by
physicians. EXPEX is devoted to the development of new theoretical
insights into this problem.

REFERENCES

Teach, R.L. and Shortliffe, E.H.: An analysis of physician attitudes
regarding computer-based clinical consultation systems. Comput.
Biomed. Res. 14:542-558, 1981.

Wallis, J.W. and Shortliffe, E.H.: Explanatory power for medical expert

systems: Studies in the representation of causal relationships for
clinical consultations. Submitted for publication, December, 1981.

E. A. Feigenbaum 294
P41 RROO785-09 Appendix C

Stanford Project: MOLGEN--AN EXPERIMENT PLANNING SYSTEM
FOR MOLECULAR GENETICS

Principal Investigators: Edward A. Feigenbaum, Ph.D.
Department of Computer Science
Stanford University

Laurence H. Kedes, M.D. (KEDES@SUMEX~AIM)
Department of Medicine

Stanford University

Stanford, California 94305

(415) 497-5897

Douglas L. Brutlag, Ph.D. (BRUTLAG@SUMEX-AIM)
Department of Biochemistry

Stanford University

Stanford, California 94305

(415) 497-6593

Contact: Dr. Peter FRIEDLAND@SUMEX-AIM
(415) 497-3728

The goal of the MOLGEN Project is to apply the techniques of
artificial intelligence to the domain of molecular biology with the aim of
providing assistance to the experimental scientist. The most substantial
problem under consideration is the task of experiment design. Two major
approaches to this problem have been explored, one which instantiates
abstracted experimental strategies with specific laboratory tools, and one
which creates plans in toto, heavily influenced by the role played by
interactions between plan steps. As part of the effort to build an
experiment design system, a knowledge representation and acquisition
package--the UNITS System, has been constructed. A large knowledge base,
containing information about nucleic acid structures, laboratory
techniques, and experiment-design strategies, has been developed using this
tool. Smaller systems, such as programs which analyze primary sequence
data for homologies and symmetries, have been built when needed.

SOFTWARE AVAILABLE ON SUMEX
Knowledge-based Experiment Design system (Friedland).
Meta-planning with Constraints experiment design system (Stefik).
UNITS system for knowledge representation and acquisition.
Interactive KORN Program for DNA sequence analysis.

GAi program for restriction map construction.

SAFE program for gene excision.

295 E. A. Feigenbaum
Appendix C P41 RROO785-09

REFERENCES

Friedland, P.E.: Knowledge-based experiment design in molecular genetics
(Ph.D. thesis). Computer Science Dept. Report, CS-79-771, Stanford
Univ.

Friedland, P.E.: Knowledge-based experiment design in molecular genetics.
Proc. Sixth IJCAI, Tokyo, August, 1979, pp. 285-287.

Stefik, M.J.: An examination of a frame-structured representation system.
Proc. Sixth IJCAI, Tokyo, August, 1979, pp. 845-852.

Stefik, M.J.: Planning with constraints (Ph.D. thesis). Computer Science
Dept. Report, Stanford Univ. (In progress)

E. A. Feigenbaum 296
P4it RROO785-09 Appendix C

Stanford Project: The MYCIN PROJECTS -- KNOWLEDGE ENGINEERING
FOR MEDICAL CONSULTATION

Principal Investigators: Bruce G. Buchanan, Ph.D.
Computer Science Department
Margaret Jacks Hall
Stanford University
Stanford, California 94305
(415) 497-0935 (BUCHANAN@SUMEX-AIM)

Edward H. Shortliffe, M.D., Ph.D.

Departments of Medicine and Computer Science
Stanford University Medical Center - Room TC117
Stanford, California 94305

(415) 497-6879 (SHORTLIFFE@SUMEX~-AIM)

Subproject Directors: MYCIN: Dr. Shortliffe
EMYCIN: Prof. Buchanan
GUIDON: Prof. Buchanan and Dr. William J. Clancey
ONCOCIN: Dr. Shortliffe and Ms. Miriam B. Bischoff

The MYCIN Projects are a set of research efforts overseen by a
collaborative group of physicians and computer scientists who are
developing intelligent systems using the techniques of knowledge
engineering. The research focus includes knowledge acquisition, inexact
reasoning, explanation, education, and the representation of time and of
expert thinking patterns. Project members currently are working in a
variety of medical domains including infectious disease therapy selection,
intelligent computer-aided instruction, and the management of cancer
chemotherapy protocols. Recent emphasis in the research has included
intensive work regarding human engineering in an effort to implement the
cancer therapy system for routine use by physicians. There is also a
heightened interest in gearing representation, knowledge acquisition, and
explanation more to the way that an expert actually thinks.

297 E. A. Feigenbaum
Appendix C P41 RROO7S85-09
MYCIN Projects (continued)

SOFTWARE AVAILABLE ON SUMEX

MYCIN-- A consultation system designed to assist physicians with the
selection of antimicrobial therapy for severe infections. It has
achieved expert level performance in formal evaluations of its ability
to select therapy for bacteremia and meningitis. Although MYCIN is no
longer the subject of an active research program, the system continues
to be available on SUMEX for demonstration purposes and as a testing
environment for other research projects.

EMYCIN-- The “essential MYCIN" system is a generalization of the MYCIN
knowledge representation and control structure. It is designed to
facilitate the development of new expert consultation systems for both
clinical and non-medical domains.

GUIDON-- A system developed for intelligent computer-aided instruction.
Although it is being developed in the context of MYCIN’s infectious
disease knowledge base, the techniques are generalizable to any EMYCIN
domain. The current research emphasis has been on an improved
understanding of how the expert thinks so as to optimize the learning
experience for the student.

ONCOCIN--This system is designed to provide oncologists with consultations
regarding the management of patients receiving chemotherapy for
cancer. Much of the knowledge in this domain is already well-
specified in protocol documents, but expert judgments also need to be
understood and modeled. Other research topics include human
engineering and the development of capabilities to help achieve
clinical acceptance of the program.

REFERENCES

van Melle, W., Shortliffe, E.H. and Buchanan, B.G.: EMYCIN: A domain-
independent system that aids in constructing knowledge-based
consultation programs. Pergamon-Infotech State of the Art Report on
Machine Intelligence, August, 1981.

Shortliffe, E.H., Scott, A.C., Bischoff, M.B., Campbell, A.B., van Melle,
W., and Jacobs, ©.D.: ONCOCIN: An expert system for oncology protocol
Management. Proc. 7th IJCAI, Vancouver, B.C., August, 1981, pp. 876-
881.

Suwa, M., Scott, A.C. and Shortliffe, E.H.: An approach to verifying
completeness and consistency in a rule-based expert system. Memo HPP-
81-5, Stanford Heuristic Programming Project; Rev. April, 1982.

Clancey, W.J. and Letsinger R.: NEOMYCIN: Reconfiguring a rule-based expert

system for application to teaching. Proc. 7th IJCAI, Vancouver, B.C.,
August, 1981.

E. A. Feigenbaum 298
P41 RROO785-09 Appendix C

Stanford Project: PROTEIN STRUCTURE MODELING (CRYSALIS)

Principal Investigator: Edward A. Feigenbaum, Ph.D.
Department of Computer Science
Stanford University

Contact: Allan TERRYGSUMEX-AIM
(415) 497-1740

The CRYSALIS system is an application of artificial intelligence
methodology to the task domain of protein crystallography. The focus is
the structure determination problem: the derivation of an atomic model of
the protein from an indistinct image of the electron density. The
crystallographer interprets these data in light of the known chemical
composition of the protein, general principles of protein chemistry, and
his own experience. The goal of the CRYSALIS Project is to integrate these
diverse sources of knowledge and data into a program that matches the
crystallographer’s level of performance in electron density map
interpretation. A successful solution to this problem must deal with
issues such as representation and management of a large knowledge base,
opportunistic reasoning, and appropriate description of the emerging
hypothesis, while keeping human engineering considerations in sight.
Automation of this task would shorten the time for protein determination by
several weeks to several months and would fill a major gap in the
construction of a fully-automated system for protein crystallography.

SOFTWARE AVAILABLE ON SUMEX

CRYSTALLOGRAPHIC DATA REDUCTION PROGRAMS (in FORTRAN):
A density map skeletonizer (SKEL37) based on an improved version
of Greer’s algorithm.
A package for locating the critical points in a map.
A general map-manipulation utility (INSPCT) that can find peaks,
display regions, and compute various statistics.

TWO LISP SYSTEMS (with the caveat that both are under active development):
A system (SEGLABELING) which heuristically parses the segmented map
into labels similar to those a crystallographer would use.
The inference system (CRYSALIS).

REFERENCES

Engelmore, R.S. and Nii, H.P.: A knowledge-based system for the
interpretation of protein x-ray crystallographic data. Heuristic
Programming Project Report HPP-77-2, Computer Science Dept., Stanford

Univ., January, 1977.

Engelmore, R. and Terry, A.: Structure and function of the CRYSALIS system.
Proc. Sixth IJCAI, Tokyo, August, 1979.

299 E. A. Feigenbaum
Appendix C P41 RROO785-09

Nii, H.P. and Feigenbaum, E.A.: Rule-based understanding of signals.
Heuristic Programming Project Report HPP-77-7, Computer Science Dept.,
Stanford Univ., April, 1977.

E. A. Feigenbaum 300