Akademik Veri Yönetim Sistemi
2021 International Conference on INnovations in Intelligent SysTems and Applications, INISTA 2021, Kocaeli, Türkiye, 25 - 27 Ağustos 2021
© 2021 IEEE.Proper dynamic network extraction is a prominent problem for timely evolving systems' modelling. This problem is seen as finding the most proper window size for extracting the most informative and least noisy time series of snapshot features. Existing solutions suffer from using only network topological properties as snapshot features, applying subjective methodologies needed user-dependent data labeling, and extracting snapshots with equal sized windows. We propose a window aggregation strategy of intelligent snapshot extraction to overcome previous issues. The idea is to collect the snapshots that already have similar link structure. Thus, the result network has fewer snapshots with different duration. Experiments on Enron and Haggle Infocomm data sets reveal that our proposal extracts more informative snapshots than using constant window sizes. Moreover, as a complementary result, it is more effective in determining proper time interval for modelling.