Expert Systems - Theories

5 Pages 1277 Words

Theories

What distinguish expert systems from other computer systems is that they give advice based on judgments. A very important part of the field therefore is the development of algorithms for reasoning under uncertainty. The tasks in the field consist of development and implementation of methods for decision support as well as construction of specific expert systems.

Semiotics is the study of signs, symbols and signification, and is therefore the study of how meaning is created, encoded and understood. Computational semiotics is understood here to be the application of semiotic theories to computer systems and interactive digital media. Three possible aspects of this are:
· The way in which meaning can be created by, encoded in, or understood by, the computer (using systems or techniques based upon semiotics).
· The way in which meaning in interactive digital media is understood by the viewer or user (again using systems or techniques based upon semiotics).
· The way in which semiotics can be used as the starting point for a system for looking critically at the content of interactive digital media.


Large-scale expert systems have found widespread use. However, developers have found that the cost of maintaining a knowledge base, over its lifetime, can be as high as the initial cost of its development. One response is to use machine learning techniques to correct the knowledge base as problems emerge. Unfortunately, standard induction methods seem ill-suited to this task, as they are designed to use training data to construct a knowledge base from scratch, and the rate at which training data is generated by field users is typically too low to support regeneration of the knowledge base each time a revision is needed. As domain knowledge is available (in the form of the knowledge base), it makes sense to take advantage of this information (even if imperfect) to bias the induction process. Techniques for theory revision...