Artificial Intelligence · Cambridge syllabus 6.3 · Lesson 2 of 3
Expert Systems
An expert system uses a knowledge base of facts and rules with an inference engine to derive conclusions. A user interface collects information and presents results. Its usefulness depends on the quality and coverage of its encoded knowledge.
By the end, you can
- Understand the purpose of a knowledge base.
- Understand the purpose of a rule base.
- Understand the inference engine.
- Understand the user interface.
- Explain how these components work together.
- Identify suitable applications of expert systems.
Components
An expert system is designed to reproduce the decision-making of a human expert in a narrow field. It is built by interviewing specialists and encoding what they know, and it is used where expertise is scarce, expensive or needed in many places at once.
It has four parts, and questions ask for all four. The knowledge base holds the facts of the subject: symptoms and diseases, minerals and their properties, faults and their causes. The rule base holds the reasoning, as conditional rules of the form if certain facts hold then a conclusion follows.
The inference engine is the working part. It takes the facts supplied about the current case, searches the rule base for rules whose conditions are met, applies them to derive new facts, and repeats until it reaches a conclusion or runs out of applicable rules. It also decides which question is most useful to ask next, which is why a good expert system asks few questions rather than all of them.
The user interface asks those questions in plain language, usually one at a time, and presents the conclusion — normally with a certainty and with an explanation of which rules led to it, so the user can judge whether to trust it.
The four together work like this: the interface collects the case facts, the inference engine matches them against the rule base using the knowledge base, and the interface reports the conclusion and the reasoning behind it.
Worth remembering
- Knowledge base: the facts. Rule base: the if-then reasoning.
- Inference engine: applies rules to the facts and chooses the next question.
- User interface: asks the questions and explains the conclusion reached.
Applications
Expert systems are used where the knowledge is deep but narrow. Medical diagnosis suggests conditions from symptoms and test results. Mineral prospecting judges the likelihood of deposits from geological survey data. Fault diagnosis works through engines, machinery, networks and software from reported symptoms. Financial systems assess loan and insurance risk; chess and strategy programs evaluate positions; help desks route problems.
What these have in common is what makes an application suitable. The knowledge belongs to one narrow field, it can be written down as facts and as rules of the form if this then that, and the system is asked about a particular case, answering questions until it can reach a conclusion.
So when a question asks whether an expert system suits a situation, look for those features. Diagnosing a fault in a washing machine from its symptoms suits one well; deciding which film a group of friends will enjoy, where there are no clear rules, does not.
Worth remembering
- Used for medical diagnosis, mineral prospecting, fault finding and risk assessment.
- Suitable where knowledge is narrow and can be written as facts and if-then rules.
- It answers about a particular case by asking questions and applying its rules.
Worked example
A troubleshooting system can apply “IF device has no power AND cable is disconnected THEN check the cable connection”.
Practice
Warm-up · not marked
Check your understanding
Answer in your own words, then compare with the worked answer. Nothing here is saved or scored.
Worked answer
It applies the stored rules to available facts to derive conclusions or recommendations.
Check your method, units and explanation against it.
Marked practice
Set 1 of 4 · easiest first · Checked as you go, with the worked answer. Sign in or create a free account to save your progress.