What Is an Expert System in IGCSE ICT?

What Is an Expert System in IGCSE ICT?

Somewhere in a stack of IGCSE ICT scripts, an examiner reads the same sentence for the fortieth time: "the computer uses AI to work out what's wrong with the patient." It's not wrong, exactly. It's also not naming a single one of the five things the mark scheme is actually looking for.

That's the trap with expert systems as a topic. Everyone already has a rough sense of what one is, roughly a program that thinks like a specialist, so answers tend to stay vague in precisely the spot where the mark scheme wants precision: the five named components, and the exact order they work in.

Quick checklist before you answer a question on this topic:

• Can you name all five components of an expert system, not just "the AI part"?

• Do you know which two components are the only ones a user actually sees?

• Can you match a given scenario to a real, named use rather than describing expert systems in the abstract?

• Can you explain why the system gives a probability rating instead of one certain answer?

• Do you know the four-step order an expert system follows to reach a diagnosis?

What Is an Expert System in IGCSE ICT?

An expert system is AI-driven software that replicates the decision-making of a human specialist. It stores factual knowledge about one narrow domain (medicine, tax, geology) plus a set of logical rules, and applies both to give advice a real expert would give, without that expert needing to be in the room.

In IGCSE ICT you're expected to define it, name its five components, and explain how it works using a worked example. Medical diagnosis is the classic one, and it's the example worth having ready first.

What Real-World Examples of Expert Systems Should You Know?

Examiners want named real-world uses, not a vague "it's like a robot doctor." The ones worth having ready:

• Medical and mechanical diagnostics (symptoms in, condition or fault out)

• Mineral exploration (interpreting geological data to locate deposits)

• Tax planning and financial assessment

• Strategy-based computer games (opponent AI making rule-based decisions)

• Identifying organisms or chemical substances

• Route optimisation for delivery fleets

If a question gives you a scenario, match it to one of these rather than describing expert systems in the abstract. That's usually what separates a full-mark answer from a half-mark one.

What Are the Five Components of an Expert System?

Five parts work together: the user interface (the only part the user sees), the inference engine (the reasoning core), the knowledge base (stored facts), the rules base (the logic connecting those facts), and the explanation system (which justifies the conclusion). Miss one of these five in an exam answer and you lose marks.

Component

What it does

User interface

The only part of the system the end user actually sees and interacts with; usually requires some training to use properly.

Explanation system

Justifies the conclusion the system reached, suggests possible next actions, and displays a probability rating for how likely each conclusion is to be correct.

Inference engine

The reasoning core. Searches the knowledge base like a specialised search tool, gathers and evaluates input data, matches it against known objects or solutions, and carries out the problem-solving.

Knowledge base

Stores the factual information and domain expertise the system draws on: the objects in that field and their attributes.

Rules base

The collection of inference rules that steers the inference engine toward a logical conclusion.

Here's a way to keep the five straight: the knowledge base is the facts, the rules base is the logic connecting them, and the inference engine applies that logic to the facts. Only the user interface and explanation system are visible to a student. The rest happens behind the scenes.

How Does an Expert System Reach a Diagnosis? A Worked Example

An expert system reaches a medical diagnosis in four steps. It asks the user a series of questions, the inference engine compares the answers against the knowledge base using the rules base, it outputs a ranked list of possible conditions with probability ratings, and the explanation system justifies the conclusion. Walk through these in order in an exam answer, in this sequence:

1. Data input: the system asks a series of questions through a touchscreen or keyboard, typically about symptoms.

2. Matching: the inference engine compares the entered symptoms against the knowledge base, applying the rules base to work out which stored conditions fit.

3. Output: the system presents a list of possible conditions, each with a probability rating, plus recommended treatment or next steps.

4. Justification: the explanation system clarifies why it reached that diagnosis, so the output isn't just a black-box guess.

Worked example to reuse in an answer: a patient enters "persistent cough and fever" via the touchscreen. The inference engine checks this against stored conditions in the knowledge base using the rules base, then returns "bronchitis: 72% probability, see a doctor if symptoms persist beyond 5 days," alongside two lower-probability alternatives, with the explanation system stating which symptoms triggered that ranking.

If you want to see this exact structure applied to different past paper scenarios, not just the medical one, the Question Bank has expert systems questions grouped by topic so you can practise matching a scenario to the right named use before it counts.

A Note on Exam Boards

The terminology above matches the conventions used in Cambridge/CIE-style IGCSE ICT (0417) teaching material. If you're on a different specification, check your board's glossary before using these five exact labels in an answer, since wording can vary slightly between boards.

For the wider hardware side of automated data capture that often comes up alongside AI topics in IGCSE ICT, see our guide to RFID and NFC.

Frequently Asked Questions

What is an expert system in simple terms? Software that copies a human specialist's decision-making by storing their knowledge as facts and rules, then applying both to a user's input to give advice, for example suggesting a likely diagnosis from a list of symptoms.

What are the five components of an expert system? The user interface, explanation system, inference engine, knowledge base, and rules base. The user interface is the only one the end user actually sees; the other four work in the background.

What is the difference between the knowledge base and the rules base? The knowledge base stores raw facts for the subject area, such as conditions and their attributes. The rules base stores the logic connecting those facts. The inference engine uses both to reach a conclusion.

Why does an expert system give a probability rating instead of a single certain answer? It's matching input against stored patterns, not confirming a fact outright, so there's rarely one guaranteed answer. A probability rating per conclusion is more honest, and the explanation system justifies the ranking.

Can an expert system make decisions outside its subject area? No. An expert system only works within the narrow domain it was built for; its knowledge base and rules base only cover that field. A medical diagnosis expert system has no rules for, say, tax planning, so it can't reason about anything outside its trained domain.

What to Actually Do With This Before the Exam

Cover the five-component table and rebuild it from memory: what each part does, and which two are actually visible to the user. Then take one scenario you haven't seen before (a farmer diagnosing a crop disease, a mechanic diagnosing an engine fault) and walk it through the same four steps as the medical example, out loud, start to finish.

If you've written a full answer explaining how a system reaches a conclusion and want to know whether it names enough of the five components to score full marks, the AI Tutor can check your reasoning against the same mark scheme logic examiners use.