CYC

CYC

CYC is an example of a frame-based representational system of knowledge, which is, in a way, the opposite of an expert system. Whereas an expert system has detailed knowledge of a very narrow domain, the developers of CYC have fed it information on over 100,000 different concepts from all fields of human knowledge. CYC also has information of over 1,000,000 different pieces of “common sense” knowledge about those concepts.

The system has over 4000 different types of links that can exist between concepts, such as inheritance, and the “is–a” relationship that we have already looked at.

The idea behind CYC was that humans function in the world mainly on the basis of a large base of knowledge built up over our lifetimes and our ancestors’ lifetimes.

By giving CYC access to this knowledge, and the ability to reason about it, they felt they would be able to come up with a system with common sense. Ultimately, they predict, the system will be built into word processors.

Then word processors will not just correct your spelling and grammar, but will also point out inconsistencies in your document.

For example, if you promise to discuss a particular subject later in your document, and then forget to do so, the system will point this out to you. They also predict that search engines and other information retrieval systems will be able to find documents even though they do not contain any of the words you entered as your query.

CYC’s knowledge is segmented into hundreds of different contexts to avoid the problem of many pieces of knowledge in the system contradicting each other.

In this way, CYC is able to know facts about Dracula and to reason about him, while also knowing that Dracula does not really exist.

CYC is able to understand analogies, and even to discover new analogies for itself, by examining the similarities in structure and content between different frames and groups of frames. CYC’s developers claim, for example, that it discovered an analogy between the concept of “family” and the concept of “country.”