Exam 15: Artificial Intelligence

Correct Answer:

Verified

Forward chaining begins with assertions and tries to match those assertions to the "if" clauses of rules, thereby generating new assertions. These may in turn be matched with "if" clauses, generating still more assertions. This is the process we used in our example. Backward chaining begins with a proposed conclusion and tries to match it with the "then" clauses of rules. It then looks at the corresponding "if" clauses and tries to match those with assertions, or with the "then" clauses of other rules. This process continues until all "if" clauses that arise have been successfully matched with assertions, in which case the proposed conclusion is justified, or until no match is possible, in which case the proposed conclusion is rejected.

Correct Answer:

Verified

1. Adequacy-The representation method must be adequate to capture all of the relevant knowledge. Because of its rich, expressive powers, a natural language representation will surely capture a lot of knowledge. However, it might be difficult to extract exactly what that knowledge is. You might have to wade through a lot of unnecessary verbiage, and you must also understand the nuances of meaning within the natural language. A formal language representation has the advantage of extracting the essentials.
2. Efficiency-We want the representational form to be minimalist, avoiding redundant information wherever possible. This means allowing some knowledge that is not explicitly represented to be inferred from the knowledge that is explicitly represented. Take, for example, that it is easy to infer from the natural language, the formal language, or the semantic net that because a given creature named Spot is a dog, he has four legs and a tail and also is a mammal and, therefore, warm-blooded. This knowledge is not captured in the pictorial format. On the other hand, it would take a much longer natural language paragraph to describe all the additional knowledge about Spot that is captured in the picture.
3. Extendability-It should be relatively easy to extend the representation to include new knowledge. For example, the semantic net can easily be extended to tack on another "dog" instance. It would also be easy to capture the fact that dogs have two eyes or that mammals do not lay eggs; these properties can simply be plugged in as new ovals connected into the network.
4. Appropriateness-The representation scheme used should be appropriate for the knowledge domain being represented. For example, a pictorial representation scheme would appear to be the most appropriate way to represent the knowledge base for a problem dealing with recognition of visual images. We saw before that a pictorial representation is probably not appropriate for the kind of knowledge about Spot that is difficult to display visually. The level of granularity needed for the intended application might also influence the appropriateness of a particular scheme. Is a given pictorial representation sufficient, or do we need to "zoom in" and expose more detail? The appropriate representational form for knowledge therefore depends on the knowledge to be captured and on the type of task for which the knowledge is to be used.