Repository | Book | Chapter

In this paper we propose to model each piece of evidence as a set of hypotheses that the evidence supports. By formalizing this idea, we can reason about knowledge based on the notion of "reliable belief". Our new understanding of knowledge highlights "reliability" of information that the agent gets from evidence. This is very different from the perspective of safe belief and its focus on "robustness". By a systematic comparison between these two kinds of beliefs, we argue that it is the reliability, not the robustness, that qualifies belief as knowledge. Finally, we explore the agent's knowledge update, particularly triggered by evidence dynamics, and present a complete dynamic logic.

Full citation: