This paper presents a fundamentally different framework for uncovering the intricate properties of evolutionary networks. Contrary to static snapshots methods, this paper first traces the timelines of the networks. Then, based on extracted smooth segments from the timelines, a graph approximation algorithm is applied to capture the frequent characteristics of the network and reduce the noise of interactions. Moreover, by employing the relationship among multi-attributes, an innovative community detection algorithm is proposed for a detailed analysis on the approximate graphs. To track these dynamic communities, this paper also introduces a community correlation and evaluation method. Finally, by applying this novel framework to several real-world networks, this paper demonstrates the critical relationship between event and social evolution, and reveals meaningful properties in actual dynamic behaviors.