Expert systems
Learn the core concepts and decision-making in Cambridge 9626 Topic 7, then practise recall, application and transfer with fictional scenarios.
Exam scope and assessment notes
This is original StudyForA teaching and fictional practice based on Cambridge syllabus outcomes, not copied syllabus text, an official question, a required software brand or a Cambridge mark scheme. Practical software tasks here are planning and reasoning exercises; StudyForA does not inspect files created in spreadsheet, database, graphics, audio/video or animation software, simulate the practical exam environment, award official marks or predict a grade. Follow the software and file-format requirements issued by your centre for your exam year.
Before you begin
- Check whether your route is AS Level (Topics 1–11) or full A Level (Topics 1–21).
Work on paper. Try each question before revealing help, and explain why your method works.
Quick readiness check
Before Topic 7: expert systems
What you will learn
- Explain user interface, inference engine, facts/rules knowledge base, explanation system and knowledge-base editor
- Trace IF…THEN rules using forward/data-driven and backward/goal-driven chaining
- Apply expert systems to diagnosis, prospecting, planning, scheduling and identification
- Evaluate consistency, speed, cost, explainability, data quality and limits of captured expertise
Rules connect known facts to possible conclusions
An expert system stores facts and IF…THEN rules in a knowledge base. The inference engine applies the rules. A user interface gathers information; an explanation system can show which rules supported an output; a knowledge-base editor lets specialists update rules. Forward chaining starts with available facts and derives results; backward chaining starts with a goal and asks which evidence would support it.
It supports a decision; it does not guarantee truth
Expert systems can make specialist guidance consistent and available, for example in fault diagnosis, plant identification, financial planning or route scheduling. Their conclusion depends on the facts, rules and domain boundary represented. New cases, missing information, bias or changed practice can make a recommendation wrong. High-impact uses need human review, tested rules and a way to handle uncertainty.
Sources for this lesson
Pause and explain
Which approach best demonstrates secure, reliable IT practice in expert systems?
Worked example
Rules: IF engine fails to start AND battery light is dim THEN possible flat battery. IF battery voltage is normal THEN check starter circuit. Trace a case with a dim light and no start.
Show the worked solution
- Enter the observed symptoms as facts.
- Match facts to the first IF condition.
- Report the rule-based possible cause and explain its uncertainty.
Answer The first rule fires because both symptoms are present, so the system reports a possible flat battery and can explain the matched rule. It is a hypothesis, not a confirmed fault. A mechanic should test battery voltage; if normal, the second rule suggests checking the starter circuit.
Try a topic test
Choose Easy, Normal or Difficult. Save your answers and open worked review after finishing.
Test this topic →From guided practice to a new situation
Write your answer and reasoning first. Use a hint only if you are stuck. The model solution is for self-checking; your written work is not automatically marked.
A plant-identification system asks a learner whether a leaf is needle-shaped. It then gives a suggestion. Name two components involved and their roles.
Give me a hint
One collects the response and another applies stored domain knowledge.
Compare my reasoning
- Identify the interface that asks for an observation.
- Identify the rules/facts and engine that match it.
- Explain how an explanation system could support trust.
The user interface asks and records whether the leaf is needle-shaped. The knowledge base stores facts and IF…THEN rules, and the inference engine applies them to the response. An explanation system could show which observations and rules led to the suggestion. The result remains limited by the accuracy of the observation and completeness of the rules.
Look for these in your work
- Names components accurately
- Traces their roles
- States a limitation
A system must test whether a particular disease is present and then request only relevant symptoms. Which chaining approach fits?
Give me a hint
Start from the conclusion to be tested, not from every possible fact.
Compare my reasoning
- Set the disease hypothesis as the goal.
- Find rules that could lead to that conclusion.
- Ask for the facts needed to test those rules.
Backward chaining is goal-driven: begin with the disease hypothesis and ask for symptoms or tests that could support or reject it. The system should allow uncertain or conflicting evidence and direct high-risk outcomes to a qualified clinician rather than presenting an unverified diagnosis.
Look for these in your work
- Identifies goal-driven reasoning
- Explains question selection
- Adds appropriate human oversight
A delivery expert system keeps recommending an old route after road closures change. What maintenance and governance are needed?
Give me a hint
The knowledge source and rules need review, then output needs validation.
Compare my reasoning
- Update route facts and rule sources.
- Test normal, boundary and closure scenarios.
- Record rule ownership, version and human override.
The knowledge base should be updated from a trusted, current route source, with rule changes reviewed by a responsible specialist. Test known open roads, closures, detours and conflicting data. Record versions and provide a human override, since stale knowledge can make a logically consistent rule system produce unsafe recommendations.
Look for these in your work
- Identifies stale knowledge
- Recommends testing
- Includes accountability
Common mistakes
- Calling the knowledge base the inference engine
- Treating a possible solution as a certainty
- Confusing forward chaining with starting from a target
What can you explain now?
A plant-identification system asks a learner whether a leaf is needle-shaped. It then gives a suggestion. Name two components involved and their roles.
Compare with the explanation
The user interface asks and records whether the leaf is needle-shaped. The knowledge base stores facts and IF…THEN rules, and the inference engine applies them to the response. An explanation system could show which observations and rules led to the suggestion. The result remains limited by the accuracy of the observation and completeness of the rules.
One collects the response and another applies stored domain knowledge.
After trying it yourself, choose your next review. This is your self-assessment.
Your review choice appears on Today. Sign in to sync it across devices.
Make it stick
What is the difference between forward and backward chaining?
If you needed the solution, close it and solve the problem again from a blank page. A correct answer is stronger when you can explain the reason for each step.