Information Technology Reference
In-Depth Information
VOLUME (OBJ1, 1)
DENSITY (OBJ1, .1)
…
• Domain Theory:
VOLUME (p1, v1)
∧
DENSITY (p1, d1)
ŗ
WEIGHT (p1, v1*d1)
WEIGHT (p1, w1)
∧
WEIGHT (p2, w2)
∧
LESS (w1, w2)
ŗ
LIGHTER (p1, p2)
ISA (p1, ENDTABLE)
ŗ
WEIGHT (p1, 5) (default)
LESS (.1, 5)
…
• Operationality Criterion: The concept definition must be expressed in terms of
predicates used to describe examples (e.g., VOLUME, COLOR, DENSITY)
or other selected, easily evaluated predicates from the domain theory
(e.g. LESS).
Determine:
• Generalization of training example is to give a sufficient concept definition for
the goal concept and to satisfy operationality criterion.
EBG system constructs explanation tree in terms of the domain theory and
makes the training example satisfy goal concept definition. Figure 9.3 shows an
explanation tree of SAFE-TO-STACK (OBJ1, OBJ2).
SAFE-TO-STACK(OBJ1, OBJ2)
LIGHTER (OBJ1, OBJ2)
WEIGHT(OBJ1, 0.1)
LESS(0.1, 5)
WEIGHT(OBJ2, 5)
VOLUME(OBJ1, 1)
ISA(OBJ2, ENDTABLE )
DENSITY (OBJ1, 0.1)
Figure 9.3. SAFE-TO-STACK (OBJ1, OBJ2) EXPLANATION STRUCTURE
Figure 9.4. Reveals generalization steps of SAFE-TO-STACK (OBJ1,
OBJ2). In term of rules in the domain theory, the goal concept expression
SAFE-TO-STACK (
) is regressed through the rule LIGHT(p1, p2)
ŗ
SAFE-TO-STACK(p1, p2); similarly, regressing LIGHTER(
x, y
x, y)
through the
rule WEIGHT (
w2)
∧
LESS (w1, w2) and the yielded
predicates are shown in the second layer. In the third layer, the conjunct
WEIGHT(
x,
w1)
∧
WEIGHT (
y,
x
,
w1)
is
in
turn
regressed
through
the
rule
VOLUME(p1,
