Expert System Architecture PDF

EXPERT SYSTEM – Architecture ESC 4506 Architecture?  The science and method of design that determine the structure of the expert system Expert system architecture  Core components:  Knowledge base  Working memory  Inference engine  Accessory components: ...

EXPERT SYSTEM – Architecture ESC 4506 Architecture?  The science and method of design that determine the structure of the expert system Expert system architecture  Core components:  Knowledge base  Working memory  Inference engine  Accessory components:  User interface  Knowledge acquisition  Explanation module EXPERT SYSTEM Core components Working Inference Knowledge Memory engine base User Explanation Knowledge interface module acquisition Accessories User Expert Definition of ES parts...  Knowledge Base(KB)  Part of an ES that contains the domain knowledge  Working Memory(WM)  Part of an ES that contains the problem facts that are generated during the session.  Inference Engine(IE)  Processor that matches the facts contained in the WM. with the domain knowledge contained in the KB, to draw conclusions about the problems....Definition of ES parts  Explanation Facility  How facility: how ES arrive to a conclusion Example: ES Conclusion: the battery is bad User: How (did you arrive to this conclusion)? ES: The Voltage is below 10 volts  Why Facility: Why the ES ask a specific question. Example: ES question: Will the car start? User: Why (do you need this information)? ES: If I know this I assume that it is an electrical problem first.  Interface (Ask question to obtain reliable information)  Tosolve expert-level problems, expert systems will need efficient access to a substantial domain knowledge base, and a reasoning mechanism to apply the knowledge to the problems they are given. Inference Engine  Is a computer program that guides the manipulation of knowledge in a knowledge base  The inferential reasoning mechanism is distinct from knowledge base  In deciding where to start a search for an answer, an inference engine can be either goal driven (backward chaining) or data driven (forward chaining) Reasoning Mechanism  Backward chaining  Involves starting with 1 @ > possible goals  Inference engine tests each goal to see whether @ not the IF clauses in the rule containing possible goal are all true ~ test each rule until an answer is found  Forward Chaining  Involves reasoning from the other direction  Starts by examining the IF clauses and searches for a solution by working from the data toward a goal @ a solution  When an answer is found to all the IF clauses in a rule containing a goal word in its THEN clause, it gives the user its recommended solutions  Can expand very rapidly Characteristics of an ES  Separates knowledge from control  Possesses expert knowledge  Reasons with symbols  Reasons heuristically  Permits inexact reasoning  Is limited to solvable problems  Works well on reasonable complexity problems  Make mistakes The Basic idea of Intelligent Problem-solving 1. Knowledge = Facts + belief + heuristics 2. Success = finding a good-enough answer with resources available 3. Search efficiency directly affects success 4. Aids to efficiency: Applicable, correct, & discriminating knowledge Rapid elimination of “blind alleys” Elimination of redundant computation Increased speed of computer operation Multiple, cooperative sources of knowledge Reasoning at varying levels of abstraction 5. Sources of increased problem difficulty: Errorful data @ knowledge Dynamically changing data The number of possibilities to evaluate Complex procedures for ruling out possibilities