An experts system has knowledge that lets it reason about its own operations plus a structure that simplifies this reasoning process. For example if an expert system is organized as sets of rules then it can easily look at the inference chains it produces to reach a conclusion. If it given special rules that tell it what to do with these inference chains it can use them to check the accuracy consistency and plausibility of its conclusions and can even devise arguments that justify or explain its reasoning. This knowledge the system has about how it reasons is called meta knowledge which just means knowledge about knowledge.
Most current expert systems have what is called an explanation facility. This is knowledge for explaining how the system arrived at its answers. Most of these explanations involve displaying the inference chains and explaining the rationale behind each rule used in the chain. The ability to examine their reasoning processes and explain their operation is one of the most innovative and important qualities of experts systems.
Self knowledge is important in an expert systems because:
a. Users tend to have more faith in the results more confidence in the system.
b. Systems development is faster since the system is easier to debug.
c. The assumptions underlying the system operation are made explicit rather than being implicit.
d. It's easier to predict and test the effect of a change on the system operations.