Graphlet decomposition of a weighted network
This work provides a method for social network analysis, but it appears incremental as it builds on existing graphlet concepts by extending them to weighted networks.
The authors tackled the problem of analyzing weighted networks by introducing graphlet decomposition to encode social information based on structure, and they developed a scalable inference algorithm combining EM with Bron-Kerbosch, demonstrating it on synthetic data, Facebook messaging patterns, and 19th-century criminal associations.
We introduce the graphlet decomposition of a weighted network, which encodes a notion of social information based on social structure. We develop a scalable inference algorithm, which combines EM with Bron-Kerbosch in a novel fashion, for estimating the parameters of the model underlying graphlets using one network sample. We explore some theoretical properties of the graphlet decomposition, including computational complexity, redundancy and expected accuracy. We demonstrate graphlets on synthetic and real data. We analyze messaging patterns on Facebook and criminal associations in the 19th century.