Expert Systems F451 AS Computing
Expert Systems - Definition Created from research into Artificial Intelligence – AI Also known as knowledge based systems. Based on Facts & Rules. A system that attempts to replicate a human expert in solving specific problems or making decisions
Expert System Components Comprises of 3 things Knowledge Base Experts in the field are interviewed. Information collected from many sources. Information is entered into the knowledge base. Inference Engine Algorithms that interrogate the knowledge base. HCI Allows anyone to use system. Must be clear and easy to use – often Natural Language
Expert System - features Limited area of expertise Rule based Can operate with uncertain data Delivers reasoned advice
Expert Systems – main components outputs inputs computer program rules facts
Expert systems - uses Medical diagnosis Fault diagnosis Geological surveys Social services Industrial uses
Expert Systems – case studies and demos ELSIE – 1990’s quantity surveying tool http://www.exsys.com/demomain.html For demos of the following expert systems, including: Cessna Diagnostic & Repair Systems Immigration and Naturalization Service Environmental Compliance Self Appraisal Weather Instrument Product Selector State of Wisconsin - Public School Disciplinary Action Advisor DriveRight Trip 500AL Hardware Installation Expert System
Expert Systems - benefits Can be faster than human experts Cost reduction over alternative diagnostics High reliability possible Consistent, rationalised decisions & results Encapsulates and distributes expertise Can be used as a repository of knowledge
Expert Systems - Limitations Can make mistakes and will not learn from them Expert knowledge is hard to obtain. Systems work best when problem is clearly defined and rules are clearly stated Can create a dearth of skills Stifles creative processes and exception handling