Cambridge A Level Information Technology (ICT): Expert systems (9626) – Revision Notes

For full explanations and worked examples, read the study guide on expert systems first. These notes are for quick recall in the final weeks.

They cover topic 7, Expert systems (section 7.1) of the Cambridge International AS & A Level Information Technology 9626 syllabus for examination in 2025, 2026 and 2027 (version 3). This is AS Level content, part of both the AS Level and the full A Level. It is examined on Paper 1 (Theory), which is based on sections 1–11; Paper 2 (Practical) candidates apply knowledge and understanding from sections 1–7.

Course hub: Cambridge A Level IT. Tick off each outcome on the 9626 checklist. Test yourself with the practice questions, or find your weak spots with a free diagnostic.

7.1 at a glance

Part of 7.1	What to recall
Possible solutions	User answers questions → inference engine applies rules to facts → possible solutions, often with probabilities → explanation available
Components	User interface; inference engine; knowledge base (database of facts + rules base); explanation system; knowledge base editor
Scenarios	Mineral prospecting; investment analysis; financial planning; insurance planning; car engine fault diagnosis; medical diagnosis; route scheduling for delivery vehicles; plant and animal identification
Chaining	Forward chaining and backward chaining, using IF…THEN rules
Data driven / goal driven	Uses in diagnoses, gaming, and AI for manipulating social media
Evaluation	Advantages and disadvantages

Definition

Expert system: a computer system that stores the knowledge of human experts in one field and uses rules to reason with it, producing possible solutions and explaining how it reached them.

Key words for full credit: knowledge of experts, rules, possible solutions (not guaranteed), explains its reasoning.

The five components

Component	What it does	Who uses it
User interface	Asks questions (often yes/no or multiple choice); accepts answers; displays conclusions and explanations	The user
Inference engine	Reasoning: matches facts to rules, fires rules, reaches conclusions, decides the next question	Works automatically
Knowledge base: database of facts	Stored facts about the field (species features, faults and symptoms)	Read by the inference engine
Knowledge base: rules base	IF…THEN rules for reasoning with the facts	Read by the inference engine
Explanation system	Shows how a conclusion was reached and why a question is asked; may give a confidence level	The user
Knowledge base editor	Adds, edits and deletes facts and rules to keep the system up to date	Knowledge engineer or experts

Must-know distinctions

• Inference engine vs explanation system: the engine does the reasoning; the explanation system reports it.
• User interface vs knowledge base editor: the user enters case data through the interface; the editor changes the stored knowledge.
• Facts vs rules: a fact states something (“species Q has smooth leaves”); a rule links conditions to a conclusion (“IF … THEN …”).

How a possible solution is produced (method in steps)

1. User interface asks a question; user answers.
2. Answer stored as a fact for this case.
3. Inference engine searches the rules base for matching rules.
4. Matching rule fires; conclusion becomes a new fact.
5. Repeat, asking further questions only where needed.
6. Display possible solutions (often ranked or with probabilities).
7. Explanation system available on request.

IF…THEN rules and chaining

	Forward chaining	Backward chaining
Also called	Data driven	Goal driven
Starts with	Known facts	A goal (hypothesis)
Works	From IF parts to THEN parts	From a THEN part back to its IF conditions
Stops when	No rule adds a new fact	Goal proved, or all goals rejected
Questions asked	Can ask for lots of data up front	Only what is needed to test the current goal

Method in steps: tracing forward chaining

1. List the starting facts.
2. Go through the rules in order. If every IF condition is a known fact and the THEN part is new, the rule fires; add its conclusion.
3. Repeat passes until a full pass adds nothing.
4. State the rules in the order they fired and the final conclusion.

Method in steps: tracing backward chaining

1. Take the goal. Find a rule whose THEN part is that goal.
2. Treat each IF condition as a sub-goal. Prove it from known facts, from another rule, or by asking the user.
3. If any condition fails, reject that rule (or goal) and try the next one.
4. Never ask the same question twice: reuse answers already given.

Worked reminder (forward chaining, investment analysis)

Fictional rules:

R1  IF time to retirement is under 10 years THEN investment horizon is short
R2  IF investment horizon is short AND attitude to risk is low THEN suggest: mainly lower-risk savings
R3  IF investment horizon is short AND attitude to risk is high THEN suggest: mix with some higher-risk funds

Facts: 7 years to retirement; attitude to risk is low. R1 fires (horizon short). R2 fires (lower-risk savings). R3 does not fire. Order: R1, R2.

Data driven and goal driven: uses

Use	Data driven (forward)	Goal driven (backward)
Diagnoses	Enter all symptoms or fault codes; system lists every condition or fault they support	Start from one suspected condition or fault; ask only for the evidence to confirm or reject it
Gaming	Character reacts to the current game state (IF player in range THEN attack)	Character has a goal and plans the steps backward to reach it
AI for manipulating social media	Activity data (likes, shares, viewing time) used to infer interests and choose content	Start from a goal (more time online, see a message) and choose content and timing to achieve it

Scenarios: one line each

Scenario	Input → possible solution
Mineral prospecting	Survey and rock sample data → probable deposit sites, where to drill
Investment analysis	Market and company data, investor aims → buy, hold or sell, with risk
Financial planning	Income, spending, debts, goals → saving, borrowing or retirement plan
Insurance planning	Age, claims history, property or vehicle → risk level, cover, premium
Car engine fault diagnosis	Symptoms, fault codes → likely fault, repair
Medical diagnosis	Symptoms, test results, history → possible conditions with probabilities, further tests
Route scheduling for delivery vehicles	Addresses, time windows, vehicles, roads → stop order and route per vehicle
Plant and animal identification	Observed features → most likely species

Identification questions should test one feature at a time, like the couplets of a classification key.

Advantages and disadvantages

Advantages	Disadvantages
Consistent; no tiredness	Errors in the knowledge base give wrong advice
Available any time, including remote areas	No common sense; fails outside its rules
Combines several experts’ knowledge	Expensive and slow to build
Explains its reasoning	Needs regular updating
Cheaper than many experts over time; trains non-experts	Over-reliance; no empathy; unclear responsibility for errors

Link every point to the scenario in the question.

Quick self-test

1. Name the two parts of the knowledge base.
2. Which component decides which question to ask next?
3. Which component would a knowledge engineer use to add a rule for a new car model?
4. Give the other name for forward chaining.
5. State what backward chaining starts with.
6. A system asks, “Why do you need my age?” Which component answers?
7. Rules: R1 IF parcel is urgent THEN priority is 1. R2 IF priority is 1 AND road is closed on route A THEN use route B. R3 IF use route B THEN schedule first stop on route B. Facts: parcel is urgent; road is closed on route A. List the rules in the order they fire using forward chaining.
8. Give one data driven use of AI in social media.
9. A doctor suspects one condition and wants to test it. Data driven or goal driven?
10. Give one disadvantage of an expert system for insurance planning.

Answers

1. Database of facts; rules base.
2. Inference engine.
3. Knowledge base editor.
4. Data driven.
5. A goal (a hypothesis or possible conclusion).
6. Explanation system (shown through the user interface).
7. R1, R2, R3: R1 gives priority 1; R2 then gives route B; R3 then schedules the first stop on route B.
8. Using a user’s likes, shares and viewing time to infer interests and select which posts or adverts to show.
9. Goal driven (backward chaining).
10. For example: if a risk rule is wrong or out of date, every customer it applies to gets an unsuitable premium; or it cannot take account of unusual circumstances that a human underwriter would notice.

Where marks are usually lost

• Naming the knowledge base without its two parts.
• Describing the explanation system as “explaining how to use the system”. It explains the reasoning.
• Saying the knowledge base editor is used to type in symptoms or answers.
• Writing that forward chaining “starts with the answer”. It starts with the data.
• In a chaining trace, stopping after the first pass when a new fact would fire another rule.
• In backward chaining, asking a question whose answer is already known.
• Describing a scenario without saying what is input and what possible solution is output.
• Giving “it is fast” or “it is expensive” with no reason and no link to the scenario.
• Saying an expert system replaces the expert. It supports a decision; it suggests possible solutions.

Next steps

• Full explanations and worked examples: expert systems study guide
• Marked exam-style questions: expert systems practice
• Related: algorithms and flowcharts (IF…THEN logic) and the digital divide
• Free diagnostics

Official syllabus

Cambridge International AS & A Level Information Technology 9626 syllabus for examination in 2025, 2026 and 2027, version 3 (published July 2025), Cambridge International, part of Cambridge University Press & Assessment. Topic 7, Expert systems, section 7.1.