Expert system — human expertise in ONE narrow area
Software built by storing the knowledge of real human experts about one narrow subject area, then using it to give advice, diagnose a problem or suggest a solution with no human expert present. It holds a knowledge base of facts, a rules base of IF...THEN rules built from them, and an inference engine that applies those rules to what the user enters. It does not reason generally — outside its one subject area it has no useful knowledge at all.
The five components and the one job of each
Knowledge base — stores the facts about the subject area, gathered from real human experts. Rules base — the IF...THEN rules built from those facts. Inference engine — matches the user's answers against the rules base, drawing on the knowledge base, to infer a possible solution. User interface — lets the user enter data or answer questions and displays the solution(s). Explanation system — shows HOW or WHY that conclusion was reached.
Seven syllabus uses, one shared structure
The syllabus names mineral prospecting, car engine fault diagnosis, medical diagnosis, chess games, financial planning, route scheduling for delivery vehicles, and plant and animal identification. Each swaps in different facts and rules but uses the same five components: data is entered, the inference engine applies the rules base to it, and possible solutions are produced — often ranked by likelihood rather than as one certain answer.
Drawn from real examiner reports.
Knowledge base ≠ inference engine
These two are the components candidates most often swap or cannot name at all, and answer spaces are frequently left blank. The split: the knowledge base stores the facts; the inference engine applies those facts and the rules base to work out — infer — a solution. "It has the answers" is too vague for either. The user interface is usually named correctly.
Flagged w22 P13 Q4 — naming the components, often left blank
Knowledge base ≠ rules base
The knowledge base is the raw facts about the subject area. The rules base is the set of IF...THEN rules built from those facts, and it is what the inference engine actually applies to the user's answers. The knowledge base is WHAT is known; the rules base is HOW that knowledge is used to reach a decision. Describing the rules base as simply "more facts" loses the mark.
Inference engine ≠ explanation system
The inference engine does the reasoning — it produces the solution by applying the rules base to the user's answers. The explanation system does no reasoning at all; it runs afterwards, showing the user how or why that solution was reached. Crediting the reasoning to the explanation system, or treating the two as one component, is a common slip.
"Thinks like a human" is not a definition
"A computer that thinks like a human" or "a system that knows a lot" scores little or nothing — neither says HOW the advice is produced. The accepted form names the mechanism: a knowledge base of facts from real human experts, a rules base built from it, and an inference engine applying those rules to what the user enters — all within ONE narrow subject area.
Expert system ≠ search engine or chatbot
An expert system is deliberately limited to one narrow subject area, is built from the knowledge of named human experts, and applies fixed IF...THEN rules. A general search engine or AI chatbot draws on very broad, unrestricted sources and is not tied to a single specialist domain. Implying an expert system can reason about anything contradicts the definition.
Route scheduling is not a satnav
Asked how an expert system schedules delivery routes, candidates describe a generic satnav journey. The expert-system answer names the process: addresses, time windows and each vehicle's load limit are entered; the inference engine applies stored rules (group nearby addresses, do not overload a vehicle) to the knowledge base; possible routes are produced and can be justified.
A benefit is not an advantage
"It is quicker" is a benefit — it compares with nothing. An advantage names what it beats: quicker than waiting for a human expert to be free. Every advantage, disadvantage, compare or discuss answer needs a comparative, plus a reason. Also name who gains — the garage, the patient, the delivery company — rather than offering a general good point.
Describe how = a sequence, not a list
Walk the process in order, naming the component at each step: the user interface asks questions, answers lead to further ones, the inference engine searches the knowledge base and applies the rules base, solutions appear with likelihoods, the explanation system justifies them.
Explain and describe need a "because"
State the point, then give the reason. "Available 24 hours" fails; "available 24 hours because it does not depend on a human expert being on duty" scores. Apply the same rule to every advantage and disadvantage you offer.
Learn the five components as a fixed list
Do not try to work out a component's job from its name — "inference" is the word candidates cannot picture, which is why those answer spaces go blank. Memorise all five with a one-line job each, so a match-the-definition question is answered straight off rather than guessed at.
Discuss = matched pairs plus a judgement
Never split the answer into a separate advantages list and disadvantages list, or into two columns — that prevents a real comparison and caps the marks. Pair each point against its counter-point, give a reason for each, and finish with a stated overall judgement.
Full notes, flashcards, Q&A and the topic quiz for every premium subject.
Premium plans are US$8.99/month or US$49.99/year — first month free.
Studying with a parent's blessing? Show them this.
