CycL

CycL is the knowledge representation language used by the Cyc project, a large AI knowledge base and inference engine developed by Cycorp. It is designed to encode background common-sense knowledge in a form that can be processed by automated reasoners. CycL expresses facts, rules, and schemas in a Lisp-like syntax and is intended to be read and written by both humans and machines.

The language supports a typed, predicate-based representation of concepts, relations, and events. Commonly used predicates include isa for type membership and genls for generalization, among others. CycL terms are typically constants that denote objects, predicates, or collections, and rules can be stated as implications that guide inference. The language supports higher-level constructs for defining relationships, constraints, and quantified statements.

Context is a central feature: knowledge is partitioned into microtheories (mts) that provide different contexts or viewpoints. Inference can be restricted to a single microtheory or allowed to combine information from multiple mts, enabling scenario-specific reasoning without corrupting other knowledge.

Semantics and use: CycL is used with Cyc’s inference engine, which applies forward and backward chaining, rules, and default reasoning to derive new knowledge. The design emphasizes interoperability with formal logical representations while accommodating the approximations needed for large-scale common-sense knowledge.

History and scope: CycL emerged with the Cyc project in the 1980s and remains the principal language for encoding Cyc knowledge. While the Cyc project continues to influence AI research and methods for knowledge-based systems, CycL is primarily used within Cyc and related tooling rather than as a general external standard.

Example: (isa Dog Mammal) (genls Dog Animal) (instance Dog Fido)