Figure 2
From: Structural measures of similarity and complementarity in complex networks
Geometric motivation and the main properties of structural similarity coefficients. (A) Metric structure induced by similarity implies transitivity of relations and the abundance of triangles. (B, C) Wedge and head triples. (D) Local clustering can be maximized even when neighbors of the focal node are very differently embedded within the network, while \(s_i\) is sensitive to this kind of non-transitivity. (E) Local closure can be maximized even for nodes with sparse 1-hop neighborhoods if they are star-like as neighbors with degree one do not generate any head triples. On the other hand, \(s_i\) is sensitive to this violation of transitivity. (F) Necessary and sufficient conditions for maximum structural similarity.
