Shell Programming in Expert Systems Applications
Artificial Intelligence Programs Use Shells to Interpret Data
Oct 3, 2009
Expert Systems are artificial intelligence programs that apply logical arguments to a knowledge-base. Prior to the use of expert systems, and other AI applications, the data in a knowledge base could only be interpreted and used by the human brain. The use of expert systems provides a means to access and utilize the information stored without the need for human involvement. These systems, which use small processes called shells, have limitations, but can provide benefits to the business world, particularly in the area of Help-desk technology.
Shells in ES Software
ES Software typically contains shells for each domain of knowledge in the knowledge-base system. For example, in a Help-Desk Expert System, one shell might encompass all of the knowledge needed to solve the problem if an Internet user called in to ask for help with non-functioning Internet connection. Within one shell is all of the information needed to troubleshoot a DSL connection. The shell also contains a user interface, that asks questions to lead the shell to the correct conclusion. In between the user interface and the stored data is an inference engine, which decides which questions are appropriate, based on information already received.
In this example, the first question asked by the shell might be, "Is the modem's power light on?" If the user responds negatively, the shell can conclude, with some degree of certainty, that the problem is due to a lack of power to the DSL modem. If the user responds that the power light is on, the shell can assume that the problem is not due to a lack of power. At that time, the shell will automatically rule out all possibilities that include power problems.
With each ensuing question and answer, the shell uses a previously identified set of rules to eliminate more potential answers, until reaching a conclusion based on the information provided and the rules set forth by the programmers.
Limitations of Expert Systems Applications
The limitations of expert systems applications lie in the input received. ES Software does not learn from experience, as a Neural Network does. An expert system receives knowledge as data input by experts, so the conclusions it reaches are only as good as the data it has been programmed with. In addition, in order to solve problems, it relies on answers and information received from a user. If the user enters inaccurate information, the expert system is unable to compensate for it. Finally, the rules set forth by the programmer influence the results reached by the shell.
An expert system is a perfect example of the phrase, "Garbage In, Garbage Out". While expert systems have limitations, they are beneficial when executed carefully. If the information provided to the program is accurate, and the rules used to reach conclusions are logical, the shells can work together to reach logical conclusions for any situation they are programmed to address.
Related Articles:
Measuring the Responses of AI Programs: A discussion of the Turing Test of Artificial Intelligence
Artificial Intelligence Myths: Myths and realities associated with current AI technology
 |
