algorithm - Graph Theory: Calculating Clustering Coefficient -
algorithm - Graph Theory: Calculating Clustering Coefficient -
i'm doing research , i've come point have calculate clustering coefficient of graph.
according this paper straight related research:
the clustering coefficient c(p) defined follows. suppose vertex v has kv neighbours; @ (kv * (kv-1)) / 2 edges can exist between them (this occurs when every neighbour of v connected every other neighbour of v). allow cv denote fraction of these allowable edges exist. define c average of cv on v
but this wikipedia article on subject says differently:
c = (number of closed triplets) / (number of connected triples)
it seems me latter more computationally expensive.
so question is: equivalent?
it should noted paper cited wikipedia article.
thanks time.
i think they're equivalent. wiki page link gives proof triples formulation equivalent fraction of possible edges formulation when calculating local clustering coefficient, i.e. calculated @ vertex. there seems need show that
sum_v lambda(v)/tau(v) = 3 x # triangles / # connected triples
where lambda(v)
number of triangles containing v, , tau(v)
number of connected triples v middle vertex, i.e. adjacent each of other 2 edges.
now each triangle gets counted 3 times in numerator of lhs. however, each connected triple counted 1 time middle vertex on lhs, denominators same.
algorithm cluster-analysis graph-theory
Comments
Post a Comment