Expert Systems & Decision Support Systems: Components, Roles & Real-World Impact

TAKEAWAYS:

• Expert Systems mimic human expertise using knowledge bases, inference engines, and rule-based reasoning to solve domain-specific problems (e.g., medical diagnosis, fraud detection).
• Decision Support Systems combine data, models, and user interaction to help managers make semi-structured decisions (e.g., sales forecasting, supply chain routing).
• Key components of ES: Knowledge base (facts/rules), inference engine (reasoning logic), user interface, and explanation subsystem.
• DSS vs. MIS/EIS: DSS supports ad-hoc analysis (e.g., "What-if?" scenarios), while MIS provides routine reports and EIS delivers strategic dashboards.
• Real-world impact: ES reduces human error in critical tasks (e.g., Ncell’s network fault diagnosis), while DSS optimizes operations (e.g., Pathao’s dynamic pricing).
• Exam focus: Define components, compare ES/DSS with other IS types, and link to Nepali case studies (banks, telecom, e-commerce).

1. Expert Systems: The Digital Expert

What is an Expert System?

An Expert System (ES) is an artificial intelligence (AI) application that emulates the decision-making ability of a human expert in a specific domain. It uses rules, facts, and logical reasoning to solve problems that typically require human expertise.

mindmap
  root((Expert System))
    Components
      Knowledge Base ["Facts + Heuristics (Rules)"]
      Inference Engine ["Forward/Backward Chaining"]
      User Interface ["Natural Language, GUI"]
      Explanation Subsystem ["Justifies Decisions"]
    Characteristics
      Domain-Specific
      Rule-Based
      Transparent Reasoning
    Applications
      Medical Diagnosis
      Fraud Detection
      Equipment Maintenance

How Expert Systems Work: A Step-by-Step Trace

1. Knowledge Acquisition: Experts provide rules (e.g., "If blood pressure > 140 AND cholesterol > 200 → Risk: High").
2. Knowledge Representation: Rules are stored in a knowledge base (e.g., CLIPS, Prolog).
3. Inference Engine: Applies rules to new data using:
   • Forward Chaining: Starts with known facts, derives conclusions (e.g., "Patient has symptoms X, Y → Disease Z").
   • Backward Chaining: Starts with a hypothesis, verifies facts (e.g., "Is it Diabetes? Check glucose levels").
4. User Interaction: System asks questions (e.g., "Does the patient have fever?") and provides conclusions.

Components of an Expert System

Advantages and Limitations

2. Decision Support Systems: The Manager’s Toolkit

What is a Decision Support System?

A Decision Support System (DSS) is an interactive IS that helps managers make semi-structured decisions (e.g., "Should we expand to Pokhara?"). It combines:

• Data (historical sales, market trends),
• Models (forecasting, optimization),
• User interface (dashboards, query tools).

flowchart TD
  A["Manager's Problem"] --> B["Data Collection"]
  B --> C["Model Selection\n(e.g., Linear Programming)"]
  C --> D["User Interaction\n(What-if? Analysis)"]
  D --> E["Decision Output\n(e.g., 'Increase budget by 15%')"]

Types of DSS

How DSS Works: A Worked Example

Scenario: Pathao wants to optimize driver earnings during Dashain.

1. Data Input: Historical ride demand, driver availability, fuel costs.
2. Model: Uses linear programming to maximize driver earnings while minimizing empty trips.
3. User Interaction: Manager asks:
   • "What if we increase surge pricing by 20%?"
   • "Which zones have the highest demand?"
4. Output: Recommends dynamic pricing zones and driver incentives.

3. Expert Systems vs. Decision Support Systems: Key Differences

4. Other Information Systems: How They Fit In

Comparison Table:

5. Real-World Applications in Nepal

Case Study 1: Nabil Bank’s Loan Approval Expert System

• Problem: Manual loan approvals were slow and error-prone.
• Solution: Developed an ES with rules like:
  • "If credit score > 700 AND income > Rs. 50,000 → Approve loan."
• Impact:
  • Reduced approval time from days to minutes.
  • Cut default rates by 15% (fewer risky loans).

Case Study 2: Daraz’s Inventory Optimization DSS

• Problem: Overstocking in some regions, stockouts in others.
• Solution: DSS with demand forecasting models predicts:
  • "Region X will need 20% more Diwali gifts this year."
• Impact:
  • Reduced inventory costs by 12%.
  • Increased sales by 8% during peak seasons.

Case Study 3: NTC’s Network Fault Diagnosis ES

• Problem: Engineers spent hours diagnosing signal drops.
• Solution: ES with rules like:
  • "If signal drops in Kathmandu → Check tower A, B, or C."
• Impact:
  • Fault resolution time dropped by 40%.
  • Reduced customer complaints during festivals.

## In the Real World

1. eSewa’s Chatbot (Expert System)

   • Idea Used: Rule-based ES for customer queries.
   • How: When you ask "Why was my payment failed?", the bot checks:
     • "Is your phone connected to the internet?" (Rule 1)
     • "Is your account balance sufficient?" (Rule 2)
   • Impact: Handles 60% of routine complaints without human agents.

2. Pathao’s Dynamic Pricing (DSS)

   • Idea Used: Model-driven DSS for surge pricing.
   • How: During Dashain, the system analyzes:
     • Real-time ride demand,
     • Driver availability,
     • Traffic data from NTC.
   • Impact: Drivers earn 30% more during peak hours.

3. Nepal Rastra Bank’s Monetary Policy DSS

   • Idea Used: Data-driven DSS for economic forecasting.
   • How: Combines:
     • Inflation rates,
     • GDP growth,
     • Global oil prices.
   • Impact: Helps set interest rates to stabilize the economy.

## Exam Tip

1. Define Clearly:

   • "An Expert System is a rule-based AI system that mimics human expertise in a specific domain."
   • "A DSS is an interactive system that supports semi-structured decisions using data, models, and user input."

2. Compare Systems:

   • Always contrast ES vs. DSS vs. MIS/EIS in a table (as above). Examiners love this!

3. Use Nepali Examples:

   • Banks: Nabil Bank’s loan ES, Global IME’s fraud detection.
   • Telecom: Ncell’s network ES, NTC’s fault diagnosis.
   • E-Commerce: Daraz’s inventory DSS, eSewa’s chatbot.

4. Trace a Worked Example:

   • For ES: Show forward/backward chaining with a simple rule (e.g., medical diagnosis).
   • For DSS: Walk through a what-if scenario (e.g., "What if Daraz reduces shipping costs by 10%?").

5. Avoid Common Mistakes:

   • ❌ "DSS is only for executives." → Wrong! Managers use DSS daily.
   • ❌ "ES can solve all problems." → Wrong! Limited to structured domains.

## Quick Revision Mindmap

mindmap
  root((Expert Systems & DSS))
    Expert System
      Definition: "AI that mimics human experts"
      Components: Knowledge Base, Inference Engine, UI
      Example: Ncell Network Diagnosis
    Decision Support System
      Definition: "Supports semi-structured decisions"
      Types: Model-Driven, Data-Driven, Document-Driven
      Example: Daraz Inventory Optimization
    Comparison
      ES: Rule-Based, Structured Problems
      DSS: Model-Based, Semi-Structured Problems
    Real-World
      Nepal: Nabil Bank, NTC, Daraz
      Global: Google’s AlphaGo (ES), Amazon’s Forecasting (DSS)