Expert Systems And The First Expert Systems

An expert system is an artificial intelligence software that makes conclusions in a way similar to human beings. There are three components to an expert system; the user interface, the inference engine and the knowledge base (which is also referred to as the rule base). The user interface of an expert system, as with all user interfaces, is the means by which the user can input data into the machine and the way in which the machine outputs its results to the user. The knowledge base is the part of the system that stores the information and facts that will be used to understand the user's inputs and come to a conclusion (an output). The third part of an expert system is the inference engine. The inference engine is the component of the system that takes the information and facts stored in the knowledge base, generates questions, receives answers and ultimately generates the conclusion. (Marshall, 2002) The subject of the facts and information stored within the knowledge base determine the area of knowledge the expert system is intelligent in, this is called the `domain` of the expert system. Knowledge bases are filled with this information by Knowledge Engineers, who are people who illicit the information from Domain Experts. Examples of domain experts would be Doctors in the case of medicine or mechanics in the case of auto repair. These facts and pieces of information can be stored in a number of ways including; rules, a decision tree or a factor table. Rules can be either forward rules or backward rules. In forward rules, you ask questions to find an outcome. In backward rules, you start with a likely outcome and try to prove it by asking questions. (Bethune, et al., 2006) Expert systems can be used for many different purposes. Generally, these purposes can be categorised under four headings; advice, diagnosis, classification and planning. The system generally regarded as being the first expert system was DENDRAL which was built in 1971 by Edward Feigenbaum at Stanford University. It was sponsored by NASA and was a classification expert system for their use on an unmanned space probe. (Encyclopedia Britanica, 2012) The probe was designed land on Mars and take soil samples which would then be analysed with a mass spectrometer. It was DENDRAL's job to take the output of the ma¬¬ss spectrometer, interpret them using the rules in the knowledge base and conclude which substances where in the soil. The domain expert in DENDRAL was chemist Joshua Lederberg. Unlike modern expert system development which almost always uses a core expert system with an empty knowledge base (a shell), Feigenbaum and his team obviously had to develop DENDRAL from the ground-up, meaning it took over twenty person-years to build. DENDRAL was, at the time, not referred to as an 'expert system' and that term was not used until 1977. (Merriam-Wester.com, n.d.)

Spurred by both the governmental and commercial success of DENDRAL (Encyclopedia Britannica, 2008), Feigenbaums's next expert system was called MYCIN. Started in 1972, just one year after DENDRAL's completion, MYCIN was an expert system used to diagnose blood diseases. It improved on his first work by adding features such as; explanation (why did it come to a conclusion), certainty factors (how sure of its conclusion was it) and, most importantly, it separated inference and knowledge, meaning future expert systems could use it as a shell with their own knowledge base. (Bethune, et al., 2006) In conclusion, an expert system is the decision making artificial intelligence pioneered by Edward Feigenbaum, the first of which was DENDRAL and the second of which was MYCIN. References Bethune, D., Kelly, T. & Liversidge, T., 2006. SQA Higher Information Systems Unit 3a: Expert Systems. Edinburgh: Heriot-Watt University. Encyclopedia Britanica, 2012. Expert System. [Online] Available at: http://www.britannica.com/EBchecked/topic/198506/expert-system [Accessed 26 September 2013]. Encyclopedia Britannica, 2008. DENDRAL. [Online] Available at: http://www.britannica.com/EBchecked/topic/745533/DENDRAL [Accessed 26 September 2013]. Marshall, A. D., 2002. Expert Systems. [Online] Available at: http://www.cs.cf.ac.uk/Dave/AI1/mycin.html [Accessed 26 September 2013]. Merriam-Wester.com, n.d. Expert System. [Online] Available at: http://www.merriam-webster.com/dictionary/expert%20system [Accessed 25 September 2013].