2 Expert SystemsAn ICT system that uses artificial intelligence (AI) to make decisions based on data supplied in the form of answers to questions
3 AI An AI is a reasoning process performed by computers to: Draw deductionsProduce new informationModify rules or write new rules
4 Neural NetworkA NN is a biological system that is used by the brain for learning new thingsIt allows ICT systems to learn by example like humansAdvantagesDisadvantagesAre good where algorithms cannot be developedOnly suited to different tasksCan identify structures in dataTime can be wasted and the system can be unpredictable if not chosen carefully
5 An Expert SystemKnowledge BaseA huge organised set of knowledge about a particular subject. It contains facts and judgemental knowledge which gives it the ability to guess like a humanA set of rules on which to make decisions (using the if-then structure). The Inference engine does reasoning by manipulating the knowledge baseThe user interface presents questions and information to the operator and also receives answers from the operatorInference EngineUser Interface
6 How to expert systems work An operator interacts with the UI and answers a series of questions.Each question gets passed to the inference engine which makes a decision depending on the reply.The inference engine looks in the knowledge base in order to find the most relevant question/answerIt is then presented to the UI.
7 How to make an Expert System Build the system from scratch using a software language suited to your taskProgram using PROLOG or ASPRINUse a piece of development software called an expert system shellThe shell contains a inference engine and user interface, but requires an expert to create a knowledge baseThis can be done with little programming skill, but knowledge of an expert system is needed
8 MYCIN – Medicine expert system Questions the user on symptoms (from a Knowledge base of 500 rules)Will ask for certain tests if neededAfter the symptoms it will advise on the best medicationCan also provide alternative medication with expected success rate
9 Advantages/disadvantages of an Expert System Consistency – they provide consistent answers for repetitive decisionsNo common sense – i.e. If someone had been shot and was bleeding, the system would still look for a causeCheaper – they are cheaper than using a human expert such as a GPCan make absurd errors if data is entered incorrectly, i.e. Weight and height entered the wrong way aroundThe expert system can consult a much larger pool of knowledge compared to a humanNot able to provide a creative response in certain situationsAvailable 24/7Not being able to realised when no answer is available to the problemThe Computer uses all the information it has, unlike a human who can forgetRelies on rules and knowledge being correct. Any mistakes may cause an incorrect diagnosis