What does backward chaining refer to in the context of expert systems?

Backward chaining is a reasoning method used in expert systems where the process begins with a hypothesis or goal and works backward to determine the necessary conditions that support this hypothesis. This approach essentially asks, "What must be true for this conclusion (hypothesis) to hold?"

When a user presents a specific outcome or conclusion, the system will look for evidence that supports that outcome, querying knowledge bases and rules to find information that justifies the hypothesis. If the evidence is insufficient, the system may ask the user for additional information or look for alternative routes to confirm or refute the hypothesis. This method contrasts with forward chaining, where reasoning starts with known facts and applies rules to infer new facts until a goal is reached. Therefore, choosing the option that defines backward chaining as beginning with a hypothesis aligns accurately with its fundamental principles in the context of expert systems.