E. Shortliffe, MYCIN: Computer-Based Medical Consultations, Elsevier, New York, 1976  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the evidence supporting the classification. It is not necessary to mention evidence against the classification or evidence pertaining to other possible classifications. For example, to explain how a particu 2 Terms appear in boldface when they are defined. lar conclusion was reached, Mycin [53] lists only the satisfied rules leading to the conclusion; it does not list the rules leading, with less confidence, to different conclusions. Finally, an explanation can be constructed by a simple transformation of the inferential path that leads to a classification. The transformation may ....

....suffice to define generalizations such as arch, stack, and large, red block. In most domains, superficial features do not suffice to define generalizations. Generalizations such as cup [61] and hammer [15] are defined in terms of function, not form. Categories such as infected by pseudomonas [53] and seedless grape [2] are generalizations defined in terms that are not readily perceivable in the context of classification. When abstract features are required to define generalizations, a gap exists between the case language and the generalization language. This gap may be bridged in two ....

E.H. Shortliffe. MYCIN: Computer-Based Medical Consultations. New York: American Elsevier, 1976.

....explanations, and analyses of computer programs in these areas, its meaning is frequently left to the reader s understanding, i.e. it is used in an implicit and intuitive manner. 1 An example of how contexts may help in AI is found in McCarthy s (constructive) criticism [42] of MYCIN [57], a program for advising physicians on treating bacterial infections of the blood and meningitis. When MYCIN is told that the patient has Chlorae Vibrio in his intestines, it would immediately recommend two weeks of tetracycline treatment and nothing else. While this would indeed do away with the ....

E. Shortliffe. MYCIN: Computer-Based Medical Consultations. Elsevier, New York, 1976.

....without representation Rodney A. Brooks MIT Artificial Intelligence Laboratory, 545 Technology Square, Rm. 836, Cambridge, MA 02139, USA Received September 1987 Brooks, R.A. Intelligence without representation, Artificial Intelligence 47 (1991) 139 159. This report describes research done at the Artificial Intelligence Laboratory of the Massachusetts Institute of Technology. Support for the research is provided in part ....

....without representation Rodney A. Brooks MIT Artificial Intelligence Laboratory, 545 Technology Square, Rm. 836, Cambridge, MA 02139, USA Received September 1987 Brooks, R.A. Intelligence without representation, Artificial Intelligence 47 (1991) 139 159. This report describes research done at the Artificial Intelligence Laboratory of the Massachusetts Institute of Technology. Support for the research is provided in part by an IBM Faculty 9 Development Award, in part by a grant from the Systems Development Foundation, in part by the ....

[Article contains additional citation context not shown here]

E.H. Shortliffe, MYCIN: Computer-Based Medical Consultations (Elsevier, New York, 1976).

....to revise past information and handle data that arrives out of chronological order. Tcs also provides a more flexible format for the specification of decision procedures than the rule based form used in vm. The tcs updating strategy, incremental update, differs from that used in the mycin program [22] (complete recalculation) but is similar to that used by the Digitalis Advisor [23] When extending a program beyond the single consultation setting of mycin, the overhead of doing an incremental update is saved by only changing that part of the reasoning affected by the change in the data. It is ....

Edward H. Shortliffe. MYCIN: Computer-based Medical Consultations. American Elsevier, New York, 1976.

....performance in the domain of mass spectrometry. The system was developed by embedding a chemist s knowledge about mass spectrometry as production rules. The knowledge base is readily extended by adding new rules corresponding to new chemical compounds. In a similar fashion, the well known MYCIN [73] project involved the formulation and encoding of informal knowledge about clinical medicine as production rules. For designers of knowledge based systems, production rules offer a representation of knowledge that is relatively easy to access and modify. This property makes them amenable to ....

....and by what representational means. Implicit specifications are provided in some rule languages by the syntactic ordering of rules and conjuncts, as in PROLOG, or the ordering of facts (by age) as in OPS5. Alternatively, formalisms have been 68 proposed making control explicit, as in MYCIN [73]. Further, how may control be syntactically specified For example, by reference to rule names or by constants matched by rules Here we consider rule systems that are executed in parallel. Thus, the typical issue of controlling problem solving is exacerbated by the need to control which paths are ....

E.H. Shortliffe. MYCIN: Computer-based Medical Consultations. Elsevier Press, New York, 1976.

....making the direct assessment of interesting probabilistic relationships feasible. One of the simplest ideas in this vein is to update the likelihoods in an existing structural probabilistic model by adjusting the conditional (and maybe the prior) probabilities based on actual data. Spiegelhalter [36] suggests indicating the model builder s confidence in a probability by expressing it as a ratio of the number of times the indicated event would occur in a number of trials. Thus, a conditional likelihood of 1 2 has the same numerical value as one of 500 1000, but shows far less confidence in the ....

Shortliffe E. MYCIN: Computer-based Medical Consultations. New York: American Elsevier, 1976.

....input, of the form: v1 : p1; v2 : p2; vn : pn, where v1 : vn are some of the values of an attribute and p1 : pn their corresponding probabilities. ffl C4.5 produces also probabilistic output, using certainty factors, similar to the ones used by some expert systems (e.g. MYCIN [Shortliffe, 1976]) ffl Soft thresholds can be used, in addition to the hard ones for numeric attributes. This feature allows for probabilistic classification near the threshold value, avoiding thus misclassifications caused by slightly erroneous input. 3.3.3 Analysis Design The basic decision tree building ....

Shortliffe, E.H. MYCIN: Computer-Based Medical Consultations. Elsevier, New York, 1976.

....in their applicability and also in the accuracy of the data. If such systems are to be successfully modelled then some mechanism for dealing with this uncertainty is required. Such a mechanism has been built and is used in many of the expert systems shells spawned from the MYCIN system (see (Shortliffe 1976)) These systems use so called confidence factors to describe the level of confidence that an item of data is correct and also the level of confidence there is in the conclusion of a rule based on its premises (see (Shortliffe 1976) or (Barr Feigenbaum 1981) for a wider discussion) These ....

.... of the expert systems shells spawned from the MYCIN system (see (Shortliffe 1976) These systems use so called confidence factors to describe the level of confidence that an item of data is correct and also the level of confidence there is in the conclusion of a rule based on its premises (see (Shortliffe 1976) or (Barr Feigenbaum 1981) for a wider discussion) These factors are used to produce derived confidence factors for data that is produced as the result of a rule firing. There is a well known, and documented, algorithm for this process but it is not discussed here since it is not directly ....

Shortliffe, E. H., ed. (1976), MYCIN: Computer-Based Medical Consultations, Elsevier, New York, USA.

No context found.

E. Shortliffe, MYCIN: Computer-Based Medical Consultations, Elsevier, New York, 1976

No context found.

Edward Shortliffe. MYCIN: Computer-Based Medical Consultations. American Elsevier, 1976.

No context found.

E. Shortliffe. MYCIN: Computer-Based Medical Consultations. Elsevier, New York, NY, 1976.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC