Default Logic

Default logic, introduced by Raymond Reiter in 1980, is a non-monotonic logic form that addresses the challenge of reasoning with incomplete or uncertain information. In classical logic, the addition of new information cannot invalidate previous conclusions, a property known as monotonicity. However, in real-world scenarios, especially in AI, new information can often change the context, making previous assumptions invalid.

Default logic provides a way to handle such situations by allowing the inclusion of default rules that apply in the absence of specific information to the contrary. These default rules are used to make inferences that are retractable when conflicting information is encountered, enabling more flexible and adaptive reasoning similar to human-like thought processes.

‍

An example application of default logic in AI is in expert systems, particularly in medical diagnosis. Consider a rule stating that, by default, a patient with a certain set of symptoms has a particular disease unless specific contrary evidence is present. Default logic allows the system to infer a probable diagnosis based on typical cases (the default assumption) but can retract that conclusion if further tests reveal additional information that contradicts the initial assumption.

Another example is in natural language processing, where default logic can help in understanding ambiguous sentences or phrases by making reasonable assumptions based on common usage or context, with the flexibility to revise those assumptions as more context becomes available.

‍