Purpose+of+an+algorithm

=Purpose of an algorithm within expert systems: for example, fault finding, product development=

By Daniel

As with any collection of data, an expert system's knowledge base may suffer from data integrity problems including data which has been correctly entered, data which has come become out of date, or data which is incomplete or missing completely Each of these problems could have a significant effect on the reliability and accuracy of the answers provided by the system. Similarly, the inference engine, like any software system may contain bugs. Erroneous or missing rules, or incorrect data processing will also affect accuracy.

The impact of these reliability problems depends on the knowledge domain. A **fault diagnosis** system which suggest an incorrect solution may cause inconvenience, but a medical system which suggest incorrect treatment could cause a much more serious health impact.

However, the key problem with expert systems is that they are dependent on the rules in their knowledge base, which cover only a small domain of knowledge, and they are unable to address problems outside of this domain, or exceptions to rules. This makes them unsuitable for some problems. Many modern AI researchers have generally moved away from developing expert systems, and towards developing systems which can learn and improve.