Describe components and structure of expert systems

7 Expert Systems – Components and Structure

Objective

Describe the components and structure of expert systems, and explain forward‑ and backward‑chaining reasoning.

7.1 What is an Expert System?

An expert system is a computer programme that mimics the decision‑making ability of a human expert in a particular domain. It captures specialised knowledge and applies reasoning techniques to solve problems that would normally require expert judgement.

7.2 Key Components (Cambridge Syllabus)

• Knowledge‑Base – Stores the domain expertise (facts, rules, heuristics, frames, semantic networks, or fuzzy sets).
• Inference Engine – Applies logical reasoning to the knowledge‑base to derive conclusions.
• User Interface – Enables the user to enter problem data and to receive the system’s output (forms, dialogue boxes, natural‑language queries).
• Explanation Facility – Shows the reasoning trace so the user can understand how a conclusion was reached (syllabus wording: “provides an explanation of the reasoning process”).
• Knowledge‑Base Editor (knowledge acquisition facility) – Tools that allow experts to input, edit and update knowledge in the knowledge‑base (e.g., interview worksheets, rule editors, machine‑learning‑assisted capture).

7.3 Structure of an Expert System

The architecture can be visualised as a layered model (top‑down):

7.4 Typical Applications (selected syllabus scenarios)

• Mineral prospecting – uses geological rules to locate ore deposits.
• Route scheduling – determines optimal routes based on distance, time windows and vehicle capacity.
• Plant/animal identification – matches observed characteristics with taxonomic rules.
• Medical diagnosis – matches symptoms to disease rules.
• Car‑engine fault diagnosis – interprets sensor data to pinpoint likely faults.
• Legal decision support – assists a solicitor in evaluating the likelihood of success for a claim based on precedent rules (non‑technical example).

7.5 Knowledge Representation Techniques

Different methods are used to encode expert knowledge:

1. Rule‑based (production) systems – IF‑THEN rules, e.g. IF temperature > 38°C THEN diagnosis = “fever”.
2. Frames – Structured records with slots for attributes and inheritance.
3. Semantic networks – Graphs of concepts linked by labelled relationships.
4. Fuzzy logic – Handles imprecise information with membership functions.

7.6 Inference Engine Mechanisms

The inference engine drives the reasoning process. Two fundamental styles are required by the syllabus:

A hybrid approach (combining forward and backward chaining) is often used in real‑world expert systems to improve efficiency.

7.7 Interaction Flow (Typical Session)

1. User enters a problem description via the User Interface.
2. The system may ask follow‑up questions to obtain additional data.
3. The Inference Engine applies reasoning (forward or backward chaining) using the Knowledge‑Base.
4. The Explanation Facility presents the reasoning path and the final solution.
5. If new knowledge is discovered, the Knowledge‑Base Editor is used to update the knowledge‑base.

7.8 Component Summary

7.9 Advantages of Expert Systems

• Preserve expert knowledge and make it widely accessible.
• Provide consistent, repeatable decisions.
• Operate 24 hours a day without fatigue.
• Facilitate training and knowledge transfer.

7.10 Limitations

• Knowledge acquisition can be time‑consuming and costly.
• Systems may lack common‑sense reasoning.
• Performance depends heavily on the quality of the knowledge‑base.
• Difficulty handling highly dynamic or ill‑structured domains.

7.11 Example: Medical Diagnosis Expert System

Consider a system that assists doctors in diagnosing respiratory illnesses. A sample rule is:

$$ \text{IF } \text{cough}= \text{dry} \;\land\; \text{fever}>38^\circ\text{C} \;\land\; \text{X‑ray}= \text{infiltrate} \;\text{THEN diagnosis}= \text{“pneumonia”} $$

The inference engine uses forward chaining to match patient data with the rule. The explanation facility then displays the diagnosis together with the justification trace for the clinician.

7.12 Classroom Activity – Mini Case Study

Prompt: Design an expert system to suggest the best fertilizer for a given soil type. Identify the five components (knowledge‑base, inference engine, user interface, explanation facility, knowledge‑base editor), propose at least three IF‑THEN rules, and decide whether forward or backward chaining would be more appropriate.

This activity reinforces the syllabus requirements for component identification, knowledge representation, and reasoning style.