On Updates of Epistemic States 

Belief Change under Incomplete Information 

In this dissertation I present some aspects of  belief-change theory and representation of knowledge as one of the main theoretical basis to formulate semantics for updates of logic programs. Firstly, there is an introduction to relevant principles and postulates, like the classical belief-revision formulation and a following proposal to make a difference between belief revision and updates. Next, there is a survey of some few proposals to update logic programs that are the main motivation for this thesis. Finally, I present a progressive approach that overrides the problems pointed out in other alternatives, and that meets most of the principles here introduced.

In particular, revising and updating knowledge bases is an important problem in knowledge representation and reasoning.. It has led to various proposals for updating logic programs, specifically with respect to the well known answer-sets semantics.. However, most of these approaches are based on the causal rejection principle, which leads to counter-intuitive behaviour. The proposed approach in this thesis is a semantics for abduction known as generalised answer sets, which allows one to choose potential models, without changing the semantics of the original given update programs. With generalised answer sets one can actually formulate semantics for updates that consist in choosing between generalised models that satisfy an intended set of properties and overcome certain problems from other approaches.. Weak Irrelevance of Syntax and Strong Consistency are two of the main properties an update semantics should manifest, which are a keystone to overcome the mentioned problems.

Finally, as an important component of logic programming and as a useful tool in the classroom, this work also provides the research community with online solver prototypes that help close the gap between theory and practice. These automatic testbeds make the semantics more accessible, and open up a path with a solid component for further more-complex prototypes of knowledge management.

Ackowledgements

The graphic above is keywords from my doctoral dissertation. The size of each word represents its frequency in the document. It was created by using Wordle.