Dynamic domination in fuzzy causal networks

Abstract

This paper presents a dynamic domination theory for fuzzy causal networks (FCN). There are three major contributions. First, we propose a new inference procedure based on dominating sets. Second, we introduce the concepts of dynamic and minimal dynamic dominating sets (DDS and MDDS) in an FCN. To reflect changes of dominance with time, we also introduce the concept of a dynamic dominating process (DDP) that has significant implications in many real-world problems. We pay a special attention to the minimal dynamic dominating process (MDDP) and develop rules for generating DDP and MDDP. Third, we investigate dynamic dominating sets with extended feedback, which we call effective dynamic dominating sets (EDDS), and related effective dynamic dominating process (EDDP). This study unveils a very important phenomenon in FCN: At any time t, either an EDDS exists or there is a dramatic change of the states of vertices. In the latter case we also identify the special structure of the sub-FCN induced by active vertices.