Extensible Spectral Clustering Method For Detecting Overlapping Communities in Large-scale Networks
Download as PDF
Hong Zhong, Deyong Jiang
The networked description of complex systems in the real world enables people to more clearly recognize and understand the functions of the system and the future development of the system, and play a guiding role in the further improvement and optimization of the system structure, improving the production efficiency of the system, and saving costs. Community structure analysis is to decompose the overall structure of the network into several communities, so that the links between nodes in the community are dense and the links between nodes in the community are sparse. Traditional clustering methods are limited by the existing computing and storage conditions, which are often time-consuming and highly dependent on storage space. Therefore, the study of overlapping community auto-detection in large data of complex network is of great significance for more scientific and rational planning of complex large data network, optimizing network structure and ensuring network service quality. Finally, this is fundamentally different from traditional data mining techniques. It is characterized by large amount of data, numerous types, wide source channels, low value density, uneven data quality, fast growth, high aging, and large variability. In this paper, a spectral clustering integration algorithm for large-scale network overlapping community discovery is proposed, because the computationally expensive spectral clustering cannot meet the needs of large-scale network community discovery.
Network overlapping community detection, spectral clustering