Posted by Ben Perry on February 11, 2001 at 20:20:55:
Okay, here is the deal now:
I have sorted out the empty S problem.
I have a new problem, though:
Is the definition of S the intersection of a clique's nodes and its direct parent's nodes, or is it the intersection of a clique's nodes and all of its parent / grandparents' nodes?
For example, the Neapolitan book (pg 281) shows that every clique's S contains only nodes that are common with its direct parent (and incidentally its grandparents).
Laura's algorithm, however, has clique 5 having elements in S that are NOT in the clique's parents, but ARE in some of the grandparents.
Is this situation okay?
------------------------------------------------------
Clq No. = 4
Clq = {smoking?, lung cancer?, bronchitis?, }
Parent = 3
Children = 5
R = {smoking?, }
S = {lung cancer?, bronchitis?, }
------------------------------------------------------Clq No. = 5
Clq = {tuberculosis?, either tub. or lung cancer?, lung cancer?, }
Parent = 4
Children = 6
R = {tuberculosis?, }
S = {either tub. or lung cancer?, lung cancer?, }
clique 4 does not have 'either tub. or lung cancer?'
clique 3 does, however. clique 3 is clique 4's parent.
-Ben Perry