Expert Systems Questions and Answers

Answer: C

A shell supplies the engine, the explanation facility, the working memory and the authoring tools, leaving only the domain rules to be written. EMYCIN was the first and CLIPS the most widely used. A knowledge base is what the shell lacks, and blackboards and truth maintenance are architectural components. The shell is the practical embodiment of separating knowledge from control.

Answer: D

CLIPS is a forward-chaining production system shell written in C at NASA, using Rete for matching and supporting rules, facts and objects. Its syntax influenced JESS and many teaching systems. Frames, logic programming and constraint solving are neighbouring paradigms that CLIPS touches without belonging to. Its portability and lack of licensing cost are why it outlived most commercial shells. Reading the definition literally, rather than recalling a formula, is the reliable route through questions of this shape.

Answer: A

The knowledge engineer interviews the expert, structures what is elicited and encodes it in the shell's formalism, mediating between two kinds of specialist. The expert supplies the knowledge and the end user runs the system. The role's difficulty, and the scarcity of people able to fill it, was one of the practical limits on how far the technology spread in the nineteen eighties.

Answer: B

Because a rule states a self-contained relation between conditions and conclusions, one can be added, removed or corrected without tracing through the others. That was the practical selling point over procedural code. File layout, variable sharing and refractoriness are unrelated to the property. Modularity is only approximate in practice, since interactions through working memory reappear as rule bases grow large. Reading the definition literally, rather than recalling a formula, is the reliable route through questions of this shape.

Answer: C

Rules interact through working memory, so adding one can change the behaviour of others in ways nobody predicted, and XCON's maintenance history is the standard illustration. Load time, memory size and question count grow far more gently. This emergent coupling is what undermined the modularity claim and motivated verification tools that look for redundant, conflicting and unreachable rules. Framing the question as which quantity the change acts upon makes the answer follow without calculation.

Answer: D

Conflicting rules draw incompatible conclusions from the same situation, so which one wins depends on the conflict resolution strategy rather than on the domain. Redundancy wastes effort without changing answers, subsumption means one rule's conditions imply another's, and circularity risks non-termination. Automatic detection of all four is the core of static knowledge base verification.

Answer: A

An unreachable rule has conditions no combination of facts and other rules can ever satisfy, so it is dead weight and usually signals an authoring error. Redundancy, circularity and conflict are the other classical defects. Finding unreachable rules requires reasoning about what the rest of the base can derive, which is why verification tools do more than compare rules pairwise.

Answer: B

Verification asks whether the knowledge base is internally sound, free of conflicts, redundancy, circularity and unreachable rules. Validation asks the different question of whether its answers are right, usually by comparing them with experts on test cases. Elicitation and maintenance are separate phases. The pairing of the two questions, building it right against building the right thing, is worth memorising in this form. Naming the cost of a choice alongside its benefit is what turns a fact into a design decision.

Answer: C

Validation runs test cases past both the system and human experts and compares the conclusions, often blind so that the evaluators do not know which is which. MYCIN's evaluation by that method is the classic instance, and it scored as well as faculty specialists. Verification, debugging and certification are related but distinct. Blind comparison remains the standard because expert disagreement itself has to be measured. Most errors here come from carrying a result across from a neighbouring case where it does not hold.

Answer: D

In the 1979 evaluation MYCIN's therapy recommendations were rated as acceptable about as often as those of infectious disease faculty, and better than those of junior staff. It was nonetheless never used clinically, for reasons of liability, integration and the labour of keeping it current. The gap between demonstrated competence and actual deployment is one of the lasting lessons of the period. The shared properties explain why these belong to one family, and the differences explain why more than one exists.