Attacking the Complexity of Prioritized Inference Preliminary Report

摘要

In the past twenty years, several theoretical models (and some implementations) for non-monotonic reasoning have been proposed. We present an analysis of a model for prioritized inference. We are interested in modeling resource-bounded agents, with limitations in memory, time, and logical ability. We list the computational bottlenecks of the model and suggest the use of some existent techniques to deal with the computational complexity. We also present an analysis of the tradeoff between formal properties and computational efficiency.