Jump to the Theo Overview page.
Jump to the Architectural Details Menu page.
Jump to the Home page.

Architectural Details

The knowledge of the system is rooted firmly in frames. The frame represent all kinds of knowledge, including problem spaces, statements of fact, and beliefs of those facts. This representation aids the indexing of knowledge to relevant problem spaces, because these frames should have the same prefix.

The architecture is also taskable because of the possible text interface between a potential user and problem spaces or learning methods. This interface is possible because the representations of the frames can be represented by text predicates, as in first order logic.

There are three main learning methods in the architecure: caching of inferred beliefs, caching explanations with learned predicates such that a truth maintenance system supported by Theo can periodically update its knowledge, and inductive inference for ordering learning methods.