Know and understand characteristics, uses and purpose of expert systems including mineral prospecting, car engine fault diagnosis, medical diagnosis, chess games, financial planning, route scheduling for delivery vehicles, plant and animal identifica

ICT Applications – Expert Systems

Learning objective

Know and understand the characteristics, uses and purpose of expert systems, with examples from mineral prospecting, car‑engine fault diagnosis, medical diagnosis, chess, financial planning, route scheduling for delivery vehicles and plant/animal identification.

1. Definition and purpose

An expert system is a computer programme that mimics the decision‑making ability of a human specialist in a narrowly defined domain. It is used to:

• provide expert advice where a human expert is unavailable or expensive,
• ensure consistent, repeatable analysis,
• speed up problem solving and reduce error,
• support training of less‑experienced staff, and
• operate remotely in hazardous or inaccessible environments.

2. Core components

• Knowledge base – stores domain knowledge as facts, rules (IF‑THEN statements) and heuristics. Certainty factors or fuzzy values may be attached to rules.
• Inference engine – applies logical reasoning to the knowledge base. It can work by:

  • forward‑chaining (data‑driven),
  • backward‑chaining (goal‑driven), or
  • a hybrid of both.

• Explanation facility – traces the reasoning path and tells the user why a conclusion was reached.
• User interface – accepts input and presents output in natural language or graphics. Best practice: guide the user to enter data in the required format and provide clear error messages.
• Learning capability (optional) – some systems can modify or extend their rule set from new cases. At IGCSE level this feature is rare and usually limited to simple rule‑base updates.

Table 1: Core characteristics of expert systems

3. Knowledge acquisition and maintenance

Acquiring expert knowledge is a critical step. Common techniques include:

• Structured interviews with domain experts,
• Questionnaires or surveys,
• Direct observation of expert work,
• Analysis of existing documents, manuals and databases.

Once the knowledge base is built, it must be maintained: rules are reviewed periodically, updated when new technology or research emerges, and deleted if they become obsolete.

4. Rule representation and handling uncertainty

Rules are normally written in IF‑THEN form. A certainty factor (CF) may be attached to indicate the confidence of the rule.

Two common ways of dealing with imprecise data are:

• Certainty factors – a numeric value between 0 (no confidence) and 1 (full confidence). The system may combine CFs from several rules to produce an overall confidence level.
• Fuzzy logic – uses linguistic terms such as “high temperature” or “moderate pressure” and maps them to a range of values (e.g., “high” = 0.7–1.0).

5. How an expert system works (process flow)

1. User enters observations through the interface.
2. The inference engine matches the data against the rule base using forward‑, backward‑ or hybrid‑chaining.
3. If a rule fires, the conclusion (with any certainty factor) is stored.
4. The explanation facility produces a trace of the reasoning path.
5. The result and explanation are presented to the user.

6. Representative examples

6.1 Mineral prospecting

Purpose: Identify areas with a high probability of valuable mineral deposits.

• Key data: rock type, magnetic anomalies, satellite imagery, historic mining records.
• Sample rule: “IF rock type = basalt AND magnetic anomaly > 150 nT THEN prospectivity = high (CF = 0.9).”
• Output: prospectivity maps with probability scores.

6.2 Car‑engine fault diagnosis

Purpose: Detect faults and suggest appropriate repairs.

• Key data: sensor readings, error codes, driver‑reported symptoms.
• Sample rule: “IF misfire = true AND fuel‑pressure < 30 kPa THEN cause = fuel‑injector fault (CF = 0.85).”
• Output: fault identification and step‑by‑step repair recommendations.

6.3 Medical diagnosis

Purpose: Assist clinicians in identifying possible diseases.

• Key data: symptoms, laboratory results, patient history, epidemiology.
• Sample rule: “IF fever > 38 °C AND rash = present AND recent travel = tropical THEN consider dengue fever (CF = 0.7).”
• Output: ranked list of likely conditions with suggested investigations.

6.4 Chess game analysis

Purpose: Recommend moves and evaluate board positions.

• Key data: current board layout, move history.
• Technique: minimax search with heuristic evaluation, end‑game tablebases.
• Output: best move, evaluation score, and a brief strategic explanation.

6.5 Financial planning

Purpose: Generate personalised investment or retirement plans.

• Key data: income, expenses, risk tolerance, tax rules.
• Sample rule: “IF age > 60 AND risk tolerance = low THEN allocate 70 % to bonds (CF = 1.0).”
• Output: asset‑allocation percentages, projected cash flows.

6.6 Route scheduling for delivery vehicles

Purpose: Optimise routes to minimise travel time and fuel cost.

• Key data: delivery addresses, traffic conditions, vehicle capacity.
• Algorithmic rule: “Apply Clarke‑Wright savings heuristic; if total distance > threshold, refine with local‑search.”
• Output: ordered list of stops, estimated arrival times, total distance.

6.7 Plant and animal identification

Purpose: Identify species from observable characteristics.

• Key data: leaf shape, flower colour, habitat, geographic range.
• Sample rule: “IF leaf shape = lanceolate AND flower colour = red THEN species = Eucalyptus camaldulensis (CF = 0.95).”
• Output: species name, classification hierarchy, conservation status.

7. Advantages of expert systems

• Available 24 hours a day – no need for a human expert on‑site.
• Consistent conclusions for identical data.
• Scalable – can serve many users simultaneously.
• Cost‑effective – reduces reliance on expensive consultancy.

8. Limitations

• Restricted to the knowledge that has been encoded; cannot cope with situations outside its domain.
• Knowledge acquisition is time‑consuming and may require specialist interview techniques.
• Lacks the intuition, creativity and common‑sense reasoning of a human expert.
• Requires regular maintenance to keep the knowledge base up‑to‑date.

9. Summary of applications