Document Type

Conference Proceeding

Publication Date



Recently, the logics of minimal knowledge and negation as failure MKNF [12] was used to introduce hybrid MKNF knowledge bases [14], a powerful formalism for combining open and closed world reasoning for the Semantic Web. We present an extension based on a new three-valued framework including an alternating fixpoint, the well-founded MKNF model. This approach, the well-founded MKNF semantics, derives its name from the very close relation to the corresponding semantics known from logic programming. We show that the well-founded MKNF model is the least model among all (three-valued) MKNF models, thus soundly approximating also the two-valued MKNF models from [14]. Furthermore, its computation yields better complexity results (up to polynomial) than the original semantics where models usually have to be guessed.


Presented at the 13th Portuguese Conference on Artificial Intelligence, Guimaraes, Portugal, December 3-7, 2007.

Attached is the unpublished, authors' version of this proceeding. The final, publisher's version can be found at