ABSTRACT
This paper is concerned with the problem of how to make inferences about an agent’s beliefs based on an observation of how that agent responded to a sequence of revision inputs over time. We collect and review some earlier results for the case where the observation is complete in the sense that (i) the logical content of all formulae appearing in the observation is known, and (ii) all revision inputs received by the agent during the observed period are recorded in the observation. Then we provide new results for the more general case where information in the observation might be distorted due to noise or some revision inputs are missing altogether. Our results are based on the assumption that the agent employs a specific, but plausible, belief revision framework when incorporating new information.
