Expert Systems

We already learned that an expert system is a computer system emulating the decision-making ability of a human expert. The tasks can have different forms - such as classifying, diagnosis, planning, etc., and also hybrid ones, i.e., tasks that combine multiple forms. Based on that we distinguish planning expert systems, diagnostic expert systems and so on. In this tutorial we focus on diagnostic expert systems only.

A diagnostic expert system works in the following way:

Knowledge base is acquired from the expert and represents general knowledge about a problem, like car diagnostics. Data of a particular problem case relate to one example, such as a particular car that is not starting. Such data can be retrieved from the user by asking, or from a database, or from sensor measurements directly etc. The core of the expert system (also called shell - "empty expert system") provides a mechanism of reasoning with the use case data using the general knowledge. It has a control mechanism that runs the reasoning, can provide an explanation system (explaining the reasoning) and maintains the knowledge about the case.

Let's jump directly to a simple example.