Springer Verlag, Lecture Notes in Computer Science, p. 41-59
DOI: 10.1007/978-3-642-28939-2_3
Full text: Unavailable
A model of agency that supposes goals are either achieved fully or not achieved at all can be a poor approximation of scenarios aris-ing from the real world. In real domains of application, goals are achieved over time. At any point, a goal has reached a certain level of satisfaction, from nothing to full (completely achieved). This paper presents a frame-work for representing partial goal satisfaction in an intelligent agent. The richer representation enables agents to reason about partial satisfaction of the goals they are pursuing or that they are considering. In contrast to prior work on partial satisfaction in the agents literature which in-vestigates partiality from a logical perspective, we propose a higher-level framework based on metric functions that represent, among other things, the progress that has been made towards achieving a goal. We present an example to illustrate the kinds of reasoning enabled on the basis of our framework for partial goal satisfaction.