06.03.2013 Views

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

A rule may also include a number of abnormal predicates. For<br />

instance, one simple rule to test the capability of car driving of X is given<br />

below.<br />

Adult (X) ∧ Car (Y) ∧ ¬ab (X) ∧ ¬ab (Y)→Can-drive (X,Y).<br />

We would now prefer to write ‘,’ instead of ‘∧’ in the L.H.S. of the<br />

‘→’ operator. Thus the above rule can be expressed as<br />

Adult (X), Car(Y), ¬ab(X), ¬ ab(Y) → Can-drive(X,Y).<br />

In this section, we would use a closed-world model for all predicates,<br />

except the ab predicates <strong>and</strong> postpone the commitment of the closed-world<br />

model for the ab predicates until we do not see other evidences [1]. We now<br />

illustrate the above principle by an example. Consider the following<br />

knowledge base.<br />

1. Tall persons are normally fast-runners.<br />

2. Diseased persons are normally not fast-runners.<br />

3. John is a tall person.<br />

4. John is diseased.<br />

The above pieces of facts <strong>and</strong> knowledge are represented below in nonmonotonic<br />

logic using ab predicates.<br />

1. Tall(X), ¬ab1(X)→ Fast-runner(X).<br />

2. Diseased (X), ¬ ab2(X) → ¬ Fast-runner(X).<br />

3. Tall (john).<br />

4. Diseased (john)<br />

Now, postponing commitment to ab1 <strong>and</strong> ab2 predicates, we by<br />

resolution principle find<br />

¬ ab1 (john) → Fast-runner (john)<br />

¬ ab2 (john) → ¬ Fast-runner (john).<br />

Now, unless we know anything about ¬ab1(john) <strong>and</strong> ¬ab2 (john), these<br />

two contingent facts are not inconsistent. In fact, we would assume ab1 <strong>and</strong><br />

ab2 to be false, unless we discover anything about them. But, it is to be noted<br />

that setting ab1 <strong>and</strong> ab2 false yields a contradiction Fast-runner (john) <strong>and</strong> ¬<br />

Fast-runner (john). To solve this problem, we have to rank the level of ab1 <strong>and</strong><br />

ab2. Suppose, we assume ab2 is more likely to be false than ab1, then we can<br />

infer ¬Fast-runner (john) is more likely to be true than Fast-runner (john).<br />

We now define a strategy for redundancy checking.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!