Incremental Partitioning of Large Time-Evolving Graphs

Many present-day datasets, such as social networks, biological networks, citation networks can essentially be modeled as Time-Evolving Graphs (TEGs). Partitioning such graphs into smaller components has applications in many diverse domains. However, utilizing the traditional static graph partitioning methods for such graphs does not meet the scale at which they evolve due to high computation cost of these methods and their long offline processing time. In this work, we approach this problem from a different perspective and present several scalable incremental algorithms for partitioning various TEGs when different modification events are applied to them. Moreover, with extensive experiments, we compare the results of heuristics in terms of different metrics to those of other mechanisms.