Visualization-Driven Graph Sampling Strategy for Exploring Large-Scale Networks

Graph sampling is crucial for analyzing and understanding large-scale networks across various domains. While numerous approaches have been proposed in the existing literature, a comprehensive evaluation of these methods, with regards to both quality and execution time, is still needed. This paper addresses this gap by offering an exhaustive review of current graph sampling techniques and by introducing three distinct modifications to the Mino-centric graph sampling (MCGS) method. These modified algorithms, along with established methods, are rigorously evaluated through a quantitative analysis that encompasses two comparative iterations and multiple metrics. In addition to the quantitative analysis, we also conduct a qualitative user study, where survey participants assess the quality of the sampling from a visual perspective. Our findings indicate that one of our modified versions of the MCGS algorithm, namely Batch major CC MCGS, not only outperforms other methods in the context of visual evaluation but also significantly optimizes execution time in comparison with the original MCGS algorithm. This improvement equips researchers and practitioners with a powerful tool for exploring large-scale networks in diverse fields.