Introduction to an Expert System in AI
Posted by johann | Filed under Artificial Intelligence, Cognitive Science
A recent paper I read had an interesting link to a research paper on Expert Systems and it’s use and place in Artificial Intelligence (AI). Below is a rudimentary introductions to the concept of Expert Systems.
It is sometimes thought of AI as being able to solve all possible problems and at other times lacking tremendously. Both these are valid cases. For the most part when dealing with Expert Systems in Artificial Intelligence, they excel because they deal with very specific and narrowly defined areas of expertise, thus the term; “Expert Systems”. Where critisism comes in is when AI is used to solve general purpose problems requiring implicit knowledge of a system or it’s parts.
In AI one of the nagging problems that has not yet been solved sufficiently is the problem of common sense. Christopher Locke wrote an interesting paper on this called Common Knowledge or Superior Ignorance.
The Common Knowledge problem can easily be explained as such. If a father has a son, then the son is younger than the father, but we also know that it is a fact that the son will always be younger than the father. The first statement can be taught to a computer with a Fact and a Rule. This would be explicit knowledge, but in order to move on, we need to understand implicit knowledge such as the the relationship of age and aging in regards to a father and son. This implicit knowledge is what makes AI for general purpose task extremely difficult.
To solve the common sense problem in regards to general purpose intelligent systems, an alternative approach was developed, thus enter Expert Systems; an attempt to mimic human performance within restricted domains of knowledge.
An Expert System is a class of computer program that can advise, analyse, categorise, communicate, consult, design, diagnose, explain, explore, forecast, form concepts, identify, interpret, justify, learn, manage, monitor, plan, present, retrieve, schedule, test or tutor.
They address problems normally thought to require human specialist for their solution. – Edward Feigenbaum
Parts of a Production Based Export System
- A basic set of rules that can be expressed as Policies, Rules and/or Guidelines
- A good user interface that makes it simple for the system to question a user and a user to respond in easily understandable terms.
- Facility for the user to question the program and a method whereby the system can learn from experience, thus giving meaning to questions and responses
- The System should also have the ability to give a reasoned explanation for conclusions that have been reached. This is not always possible and the information represented by the system could be probabilistic and as such an Expert System should offer various possible answers with the likelihood of each.
- The user should then be able to question the system as to how each conclusion was reached and the likelihood calculated.
- The system can then respond with an account of the steps taken to reach the conclusions and likelihood.
Intellectual Athletic ex-pat South African American Vegetarian Techie who loves his wife, parents and all creatures regardless of their current state of delusion.