The clustering degree algorithm (CDA; Wang et al. 2018) scores a node by its clustering degree \(CD_i = [\alpha d_i + (1-\alpha) s_i] / [1 + \exp(-C_i^w)]\) plus the clustering degrees of its neighbors, each scaled by the edge weight: $$PC_i = CD_i + \sum_{j \in N(i)} \frac{w_{ij}}{w_{\max}} CD_j.$$ Here \(d_i\) is the degree, \(s_i\) the strength, \(C_i^w\) the weighted clustering coefficient of Barrat et al., and \(w_{\max}\) the largest edge weight in the network.
Arguments
- x
Network input accepted by
centrality.- cda_alpha
Weight \(\alpha\) of the degree against the strength, between 0 and 1. Default 0.5, the value of Wang et al. (2018). A value outside that range raises an error.
- ...
Further arguments to
centrality. The measure usesweighted(defaultTRUE) andnormalized(defaultFALSE).
Details
The measure works on an undirected weighted network. On a directed
network the weights of the two arcs between a pair are added, and
weighted = FALSE uses the simple undirected skeleton. Self-loops
are removed, and mode and invert_weights have no effect.
A node with fewer than two neighbors has clustering 0, and an isolated
node scores 0. Negative or non-finite weights raise an error. With
cda_alpha = 1 the scores are unchanged when every weight is
multiplied by a constant, and with cda_alpha = 0 they scale by
that constant. On an unweighted network cda_alpha has no effect.
References
Wang, Q., Ren, J., Wang, Y., Zhang, B., Cheng, Y., & Zhao, X. (2018). CDA: A Clustering Degree Based Influential Spreader Identification Algorithm in Weighted Complex Network. IEEE Access, 6, 19550-19559. doi:10.1109/ACCESS.2018.2822844 .