关键词: 大数据, 社交网络, 社区聚类, 随机游走抽样

Abstract: In the era of the digital economy, social networks as an important carrier of the digital platform economy have been widely studied by scholars at home and abroad. The commercial application value of the social network is huge in the context of big data, but because of its unprecedented network scale, the traditional network analysis method is no longer applicable at a high computational cost. Using a network sampling algorithm to obtain the sample network to infer the whole network can save computing resources. Therefore, the quality of the sampling algorithm will directly affect the accuracy of social network analysis conclusions, but the existing social network sampling algorithm is prone to ignoring the network internal topology and falling into the local network, low sampling efficiency, and other defects. To make up for such defects we propose a clustering random walk sampling algorithm based on the community characteristics of social networks with big data. This method first uses the community clustering algorithm to divide the original network nodes into communities to obtain multiple community networks, and then conducts random walk sampling for each community to obtain sample networks for inference. The results of numerical simulation and case application show that the proposed algorithm outperforms the traditional methods and can better retain the structural features of the original network while reducing the network size. In addition, the algorithm can also carry out parallel operations to promote sampling efficiency, which has great practical significance for the sampling of large-scale social networks in the context of big data.

Key words: Big Data, Social Network, Community Clustering, Random Walk Sampling