Mycin is on page 68 of coursenotes

R1 is on page 98 of coursenotes

Mycin some in section 8-2: certainty factors

(me: see 222:History of ES)

Also see Chapter 20

R1

Task: design

Domain: dec

Control: forward

Generate and Test vs. Match

exhaustive acceptable

chooses correct rule

(which rules preconditions are satisfied)

Production system:

production memory ... rules

working memory

inference engine: recognize-act

rules viewed as 'state transition operators'

(builds larger&larger partial configurations)

DataBase: frames

Conflict Resolution:

fan-in, fan-out Problems

MYCIN - Diagnosis and Treatment for Infectious Diseases [Shortliffe, 1976]

TEIRESAIS - metarules; knowledge acquisition

Meta-rule 001

If (1) the infection is a pelvic-abscess, and

(2) there are rules that mention in their premise Enterobacteriaceae, and

(3) there are rules that mention in their premise gram

positive rods

Then There is suggestive evidence (.4) that the rules dealing with Enterobacteriaceae should be envoked before those dealing with gram positive rods

Pg.551

[Davis, doctoral diss., 1976] ...supported explantion& acquisition

GUIDON - teaching diagnostic problem-solving (tutorial) ICAI: Intelligent Computer Aided Instruction [Clancey, doctoral diss.,1979]

EMYCIN - domain independent Mycin [VanMelle, doctoral diss., 1980 ]

NEOMYCIN - domain independent Teiresias - separates diagnostic strategy from domain knowledge.

-

Uses hierarchical organization of data and hypothesis (makes implicit design knowledge explicit) [Clancey & Letsinger, 1981]

Generic Types (categories of applications):

arguments:

primitives - what are tasks and what are methodologies

- atomic vs. molecular types

eg. is "analogy" a task or method?

is "consultation" (MYCIN) a category?

Interpretation vs. Diagnosis

(difference in problem-type?)

Interpretation vs. Perception

Diagnosis and Classification

- explicit properties (due to the problems representation) vs. emergent properties (due to implicit aspects of an agent's strategy)

* significance - explanatory power

Significant Issue: Task vs. Domain

Issue not raised:

significance in both: What a system does AND How it does it

SOPHIE: uses multiple representations of knowledge (e.g., simulation model and semantic net)

WUMPUS & GUIDON& CSRL: use representations that can be both interpreted and used to generate teaching text (e.g., rules...?)

SCHOLAR: uses network representations of knowledge that capture "importance"

WUMPUS: representations that capture "complexity", "prerequisite" associations, "analogy" and "generalization" relations

WEST & BUGGY: representations that allow for variants on expert performance (for modeling the student)

PIP: Issues of short-term and long-term memory Cognitive processes: a network of frames

ABEL: Causal network, levels of detail

KNOBS: Constraint satisfaction and propogation

DART: hierarchical design models

See Handbook of Artificial Intelligence for more info