Network backbone

Do networks have distinct “backbones” or “highways” that constitute the essential structure? (vandenHeuvel2012highcost) One can think of it as a data filtering or visualization method as well. Many weighted networks are too dense to visualize/comprehend. So what would be the ways to prune less important edges and nodes?

There is a family of methods that extract the most salient edges (usually) based on the distribution of weights. the definition of ‘saliency’ can vary. For instance, Mercier2022effective proposes a method to sparsify a network based on the preservation of epidemic dynamics.

• Googling Social Interactions: Web Search Engine Based Social Network Construction
• A tool for filtering information in complex systems
• Extracting the multiscale backbone of complex weighted networks
  • https://gist.github.com/bagrow/11181518 (code)
• Information filtering in complex weighted networks
• Nonparametric Sparsification of Complex Multiscale Networks
• link salience
  • Robust classification of salient links in complex networks
  • The Structure of Borders in a Small World
  • Parameter-free identification of salient features in complex networks
• Unwinding the hairball graph: pruning algorithms for weighted complex networks
• Irreducible network backbones: unbiased graph filtering via maximum entropy
• A Pólya urn approach to information filtering in complex networks

We can also make use of the underlying bipartite structure to extract hierarchical structure.

• Inferring monopartite projections of bipartite networks: an entropy-based approach
• Extracting Tag Hierarchies
• Extracting hierarchical backbones from bipartite networks

See also Uncovering hidden dependency in weighted networks via information entropy.