Abstract: Distributed flexible resources are showing a trend of sea quantization. In this context, there is an urgent need to develop efficient resource organization techniques adapted to the trend. Following the idea of Internet resource organization, this paper proposes a self-organization model and its overall architecture for massive flexible resources based on the social network. Specifically, we first design interactive attraction mechanisms for participating entities adaptable to self-organization, including individuals outside the group, individuals inside the group, and group masters. Next, self-portrait and hetero-portrait techniques of resource individuals, self-organizing groups, and grid companies are proposed. Then, we present a data-driven dynamic modeling approach for the self-organizing evolutionary process. Finally, we suggest a guided evolution technique for self-organization tailored to typical grid demands. Additionally, a self-organizing simulation system based on NetLogo is constructed, and the simulation results validate the effectiveness of the proposed architecture and technologies.