Total Knowledge

Knowledge Representation and Reasoning

Volume

Proceedings of the AAAI Conference on Artificial Intelligence, 17

Nonmonotonic Reasoning

In this paper, we analyse a concept of total knowledge based on the idea that an agent’s total knowledge is the strongest proposition the agent knows. We propose semantics for propositional and first-order languages with a modal operator TK representing total knowledge, and establish a result showing that total knowledge is epistemically categorical, in the sense that it determines the agent’s knowledge over a broad range of contents. We show that (subject to some restrictions) total knowledge is always total knowledge of an objective content, and that, for such objective contents, our TK-operator corresponds in a straightforward way to Levesque’s operator O.