An Influence Diagram Approach for Multiagent Time-Critical Dynamic Decision Modeling Le Sun 1 , Yifeng Zeng 2 and Yanping Xiang 1 1 School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 610000, China {sunle2009, xiangyanping}@gmail.com 2 Dept. of Computer Science, Aalborg University, DK-9220 Aalborg, Denmark yfzeng@cs.aau.dk Abstract. Recent interests in multiagent dynamic decision modeling in partially observable multiagent environments have led to the development of several representation and inference methods. However, these methods have limited application under time-critical conditions where a trade-off between model quality and computational tractability is essential. We present a formal representation for modeling time-critical multiagent dynamic decision problems through interactive dynamic influence diagrams. The proposed model, called interactive time-critical dynamic influence diagrams, has the ability to represent space-temporal abstraction in multiagent dynamic decision models. More importantly, we take the notion of object-orientation design and make the representation flexible and reusable. The new design facilitates the modeling and implementation of models’ self-expansion and self-compression. Keywords: Time-Critical Decision Making, Multiagent Systems, Model Construction. 1 Introduction Timely action is often critical in facing rapid changes in the real world. The time- critical dynamic decision problem is to decide or select a course of actions that shall achieve a set of goals while they must be executed under time constraints. It may be considered as a real-time decision problem[1] that seeks an optimal trade-off between solution quality and solution time via the use of the most appropriate model and solution algorithm. There is a growing line of interest, mainly on a single-agent setting, for addressing time-critical dynamic decision problems [2], [3], [4]. Most of previous work adopts a type of normative systems, e.g. Bayesian networks and influence diagrams [2]. Recently, Xiang and Poh [3], [5] proposed a formal representation of time-critical dynamic influence diagrams that provide explicit support for the modeling temporal processes and dealing with time-critical situations. Their work is applicable in a single-agent decision domain. Time-critical decision modeling is more significant for multiagent applications due to the complicated decision process and solutions. The modeling of time plays an