🧠 Rule-Based Expert Systems

1. 🎯 Learning Objectives

• Trace the Match-Resolve-Execute cycle
• Apply Specificity, Recency, and Refractoriness for conflict resolution
• Explain Rete net construction and token propagation

2. 📖 Core Content

3.1 Architecture

• Knowledge Base: IF-THEN rules
• Working Memory: Current facts
• Inference Engine: Match-Resolve-Execute cycle

3.2 Match-Resolve-Execute Cycle

text
repeat:
    // MATCH
    conflict_set = [rule for rule in KB if all premises match WM]
    // RESOLVE
    selected = resolve(conflict_set)
    if none: break
    // EXECUTE
    add conclusions of selected to WM

3.3 Conflict Resolution

3.4 Rete Algorithm

• Alpha nodes: Test single conditions
• Beta nodes: Join conditions (merge two matches)
• Terminal nodes: Complete rule matched Token propagation: When a fact enters the network, it flows through matching alpha nodes, gets stored in beta memories, and triggers terminal nodes when all conditions match.

4. 📝 Practice Questions

Q1: Rules: R1(A->X), R2(A&B->Y), R3(A&C->Z). WM: A(old), B(new), C(older). Which fires with Specificity? With Recency?

Answer: Specificity: R2 and R3 tie (both 2 conditions), R1 loses. Recency: R2 matches newest fact (B), so R2 fires. Join Discord PreviousGraphplan & PSPNextCSP & AC-3