—The analysis of social networks is concentrated especially on uncovering hidden relations and properties of network members (vertices). Most of the current approaches are focused mainly on different network types and different network coefficients. On one hand, the analysis can be relatively simple; on the other hand some complex approaches to network dynamics can be used. This paper introduces a novel aspect of network analysis based on the so-called Forgetting Curve. For network vertices and edges, we define two coefficients, which describe their role in the network depending on their long-term behavior. Using one of these parameters we reduce the network to smaller components. We provide some experimental results using DBLP1 dataset. Our research illustrates the usefulness of the proposed approach. Keywords-social network reduction, memory, stability, complexity reduction, co-authorship network, visualization I. MOTIVATION Generally, the term memory is understood as committing...