Abstract

Speeding up inferences made from large knowledge bases is a key to scaling up knowledge based systems. To do so, a system must have the ability to automatically identify and ignore information that is irrelevant to a specific task. Identifying irrelevant knowledge is also key to enabling reasoning in environments in which several systems (and their respective knowledge bases) interoperate. This dissertation considers the problem of reasoning about irrelevance of knowledge in a principled and efficient manner. Specifically, it is concerned with two key problems: (1) developing algorithms for automatically deciding what parts of a knowledge base are irrelevant to a query and (2) the utility of relevance reasoning. As a basis for addressing these problems, we present a formal framework for analyzing irrelevance. The framework includes a space of possible definitions of irrelevance, based on a proof theoretic analysis of the notion. Within the space of definitions, we identify the class of...

Citations