How Expert Systems Work: An Overview

Expert Systems are a class of computer intelligent information systems with a large amount of specialized knowledge. They use specific domain expert knowledge and reasoning techniques in artificial intelligence to solve and simulate various complex and specific problems that usually require human experts, achieving a level of problem-solving equivalent to that of experts. They allow experts’ expertise to be utilized regardless of time and space, maximizing their effectiveness.

Expert systems consist of knowledge base, knowledge acquisition, human-computer interaction interface, integrated database, inference engine, and interpreter. Below we will introduce each component and its working principle:

Expert system composition principle

Building a knowledge base is a necessary step before an expert system can be put into use. The knowledge base is where expert knowledge is stored, and the quantity and quality of knowledge determine the quality of the expert system. The knowledge base contains various rules, the most common being production rules, which express knowledge rules in the form of “If (premise), Then (result)” for actions to be executed when certain conditions are met. The knowledge stored in the knowledge base includes two types: one is the knowledge itself, such as concepts and relationships between knowledge; the other is the unique experience rules, judgments, and intuitions of human experts.

The construction of the knowledge base can be achieved through the knowledge acquisition module, which refers to the process of obtaining knowledge from knowledge sources to build the knowledge base. Expert systems rely on effective interaction between domain experts and knowledge engineers to acquire knowledge. Domain experts collect and organize knowledge, while knowledge engineers convert knowledge into a symbolic form suitable for system use, that is, algorithmic representation of knowledge, while determining the logical relationships and reasoning rules among knowledge. Therefore, the process of machine cognition is to process input data into output symbols according to algorithmic rules, forming knowledge representation.

After building the knowledge base, users can start using the expert system. The human-computer interaction interface is the interface through which the expert system interacts with users—users, as well as experts who train the expert system and knowledge engineers, can achieve human-computer dialogue through the human-computer interaction interface.

The content of the dialogue is temporarily stored in the integrated database, including the raw data, intermediate results, and final conclusions required during the inference process. It can be seen as a blackboard for discussing solutions, temporarily recording the machine’s “thought” process.

The reasoning process of the machine is completed in the inference engine, which is the bridge connecting user needs and the knowledge base, and is also the core of the system’s intelligent design. It is akin to the brain of an expert, thinking about problems and solving them. First, the inference engine retrieves the user’s inquiry from the integrated database; then it looks for relevant rules in the knowledge base and draws conclusions based on those rules; finally, it stores the conclusions that match the rules in the integrated database.

Working Principle of the Inference Engine

After obtaining conclusions, it is also necessary to convert them from machine language to natural language and present them to different users through the human-computer interaction interface. The interpreter plays this role; it acts as a translator between the user and the expert system. Additionally, it provides different answers based on the varying focuses of different users on knowledge. For example, the interpreter can output high-quality solutions to domain experts; for students in the domain, it provides detailed processes for problem-solving.