This paper proposed a framework and an algorithm for identifying communities in dynamic social networks. In order to handle the drawbacks of traditional approaches for social network analysis, we utilize the community similarities and infrequent change of community members combined with community structure optimization to develop a Group-based social community identification model to analyze the change of social interaction network with multiple time steps. According to this model ,we introduced a greed-cut algorithm and depth-search-first approach and combine them to develop a new algorithm for dynamic social interaction network recognition (called ADSIN). In addition, we conduct experiments on the dataset of Southern Women, the experiment results validate the accuracy and effectiveness of ADSIN.

No abstract

A universal, repeatable, and controllable integration of single-mode optical fiber and mixed-phase TiO2 is used to manufacture a compact fiber-integrated scattering device. Based on the device, we achieve a high-performance and compact fiber-based speckle spectrometer, which has a resolution of 20 pm over a bandwidth of 15 nm, in the 1550 nm range. We test the capability of our proposed spectrometer to reconstruct narrow linewidth and broadband optical spectrums, and compare the performance with that of a traditional optical spectrum analyzer.