Research On Sampling And Visualization Of Complex Network

In recent years,various complex networks have gradually become an inseparable part of people’s life.However,due to the massiveness and unreachability of a complete complex network,it has been a great concern to get a representative sample from the original network,in order to estimate the original network’s properties or to replace it as an inexpensive object of research.In this paper,we firstly conduct several experiments to compare the efficiency and accuracy of 6 popular sampling methods including random sampling,graph traversal sampling and random walk sampling.Based on the comparison,we give suggestions on how to choose proper sampling methods.Secondly,we combine multi-way sampling and interval sampling to improve traditional sampling methods in a cross-sampling way,thus greatly reducing repetition rate and increasing efficiency.Moreover,we propose bi-graph sampling methodology to deal with several problems encountered when applying traditional way to sample directed network,which also shows higher efficiency than other sampling methods.Finally,we construct a network sampling visualization system to better display the process and samples of various sampling methods.