Expert System Architecture Overview

Expert System Architecture

• Expert systems aim to mimic human expert decision-making.
• Architecture is the science and method to design structure of expert systems.
• Core components include knowledge base, working memory, and inference engine.
• Accessory components include user interface, knowledge acquisition, and explanation module.

Core Components

• Knowledge Base (KB): Stores domain-specific knowledge. Contains facts, rules, and heuristics.
• Working Memory (WM): Stores temporary information during problem-solving. Holds facts relevant to the current problem.
• Inference Engine (IE): Processes information from the KB and WM. Derives conclusions to solve problems. This engine applies rules to find solutions.

Accessory Components

• User Interface: Enables interaction between user and the system.
• Knowledge Acquisition: Acquires domain knowledge from human experts. This is crucial for the accuracy and usefulness of the system.
• Explanation Facility: Explains the system's reasoning process and conclusions. This helps the user understand the system's decisions.

Expert System Definitions

• Knowledge Base (KB): Contains general knowledge.
• Working Memory (WM): Stores facts gathered by inference.
• Inference Engine (IE): Analyzes the KB and WM to reach conclusions. Uses rules and facts to reach a conclusion. This is the "reasoning" component.

Explanation Facilities

• How Facility: Shows how the system arrived at a specific conclusion, often illustrating how a conclusion is reached.
• Why Facility: Explains why a specific question was asked. Shows the relationship between the question and problem-solving strategy.

General aspects

• Expert systems need a substantial knowledge base for solving complex problems.
• They rely on efficient access to and application of this knowledge.

Inference Engine

• A computer program that directs the use of a knowledge base.
• Separates knowledge from the control process of problem-solving in the knowledge base.
• Can handle decision-making through goal-driven (backward chaining) or data-driven (forward chaining) methods.

Reasoning Mechanisms

• Backward Chaining: Starts with a potential solution (goal). The system checks the knowledge base for rules relevant to the goal. Continues until a solution or contradiction is found.
• Forward Chaining: Starts with known facts and rules. The system applies rules based on the facts until it reaches a goal or solution.

Characteristics of Expert Systems

• Separates knowledge from the control process.
• Possesses expert knowledge.
• Reasons using symbols and heuristics.
• Permits inexact reasoning.
• Limited to solvable problems.
• Works well with problems that have a reasonable level of complexity.
• Can sometimes make mistakes.

Intelligent Problem Solving

• Key to success is efficient access and application of knowledge.
• Knowledge includes facts, beliefs, and heuristics.
• A successful problem-solving process depends significantly on how well knowledge is structured and accessed.
• Factors like the quality of data, the number of potential solutions, and the complexity of procedures also affect problem solving.