This model extends the popular eigenvector-based measure by considering centralities as latent attributes that are constrained by the eigenvector centrality relation. By treating centrality estimation as probabilistic latent variable inference, this approach directly addresses the problem of uncertainty when inferring node importance and permits principled assimilation of repeated weight observations.
- Strongly-connected network
- With non-negative weights
- Soh, H. 2014. Probabilistic Network Metrics: Variational Bayesian Network Centrality. arXiv preprint arXiv:1409.4141.
- SOH, H. 2014. Variational Bayesian Network Centrality. arXiv preprint arXiv:1409.4141.
There are no comment yet.