Rule-Based Reasoning and Expert Systems

What Is a Rule-Based Reasoning System?

A rule-based reasoning system stores facts and applies logical rules to derive conclusions or recommend actions.

For example, a technical-support system might use rules to identify possible causes of a network problem based on symptoms reported by the user.

A rule-based system typically contains:

• Knowledge base: Facts and rules.
• Inference engine: Applies the rules to available facts.
• User interface: Collects information and presents results.

Some systems also maintain an explanation component that describes which facts and rules contributed to a conclusion.

What Is an Expert System?

An expert system is an AI application designed to solve problems in a specialized domain by applying represented knowledge and reasoning techniques.

Traditional expert systems often rely on rules supplied by domain experts.

Examples include:

• Equipment troubleshooting.
• Configuration validation.
• Policy eligibility checks.
• Technical support.
• Certain forms of decision support.

An expert system does not automatically possess human expertise. Its results depend on the quality and completeness of the knowledge it contains.

Example: A Basic Troubleshooting System

The following Python program uses a few predefined rules to suggest a possible next step when a website cannot be accessed.

def troubleshoot_website(internet_available, site_available):
if not internet_available:
return "Check the network connection"
if not site_available:
return "Check the website or server status"
return "Check browser settings or other causes"
print(troubleshoot_website(False, False))
print(troubleshoot_website(True, False))
print(troubleshoot_website(True, True))

Output:

Check the network connection

Check the website or server status

Check browser settings or other causes

The program uses the input conditions to select a recommendation. It does not independently test the network or verify the server's status.

Forward Chaining

Forward chaining begins with known facts and applies rules to derive new facts.

Example:

• Fact: The order is paid.
• Fact: Stock is available.
• Rule: If the order is paid and stock is available, the order can proceed to fulfillment.
• Conclusion: The order can proceed to fulfillment.

This approach is useful when the system receives new facts and needs to determine what follows from them.

Backward Chaining

Backward chaining begins with a goal or query and works backward to determine which facts or rules would establish it.

Example:

Goal: Can the order proceed to fulfillment?

The system checks whether:

• Payment is complete.
• Stock is available.

If both conditions are established, the goal is supported.

Forward chaining asks, "What conclusions follow from the facts I have?"

Backward chaining asks, "What facts do I need to establish this goal?"

Forward vs. Backward Chaining

Limitations of Rule-Based Systems

• Rules must be created, tested, and maintained.
• Large rule sets can become difficult to manage.
• Conflicting rules may produce inconsistent results.
• Missing knowledge can prevent the system from reaching a useful conclusion.
• Traditional rule-based systems do not automatically learn new rules from data.

Practical Exercise

Design a small rule-based system for an online store.

Requirements:

1. Store facts about whether an order is paid and whether stock is available.
2. Create a rule for determining whether an order can proceed to fulfillment.
3. Add a rule that flags unpaid orders.
4. Test the system with different combinations of facts.
5. Explain which rules produced each result.

Important: In a real e-commerce system, order processing should also account for fraud checks, payment confirmation, inventory consistency, and other business requirements.