Link prediction in complex networks: A survey
Citations
1,553 citations
1,452 citations
980 citations
Cites background from "Link prediction in complex networks..."
...It is shown that high-order heuristics often perform better than first-order and second-order heuristics [6]....
[...]
...A large h enables learning more global graph structures which is desirable since global heuristics such as PageRank and Katz are shown to have better performance than local ones such as common neighbors [6]....
[...]
...These scores measure the proximity or connectivity of two nodes [1, 6]....
[...]
967 citations
Cites background from "Link prediction in complex networks..."
...In recent years, researchers have extensively studied on many important machine learning applications in networks, such as vertex classification [Sen et al., 2008], tag recommendation [Tu et al., 2014], anomaly detection [Bhuyan et al., 2014] and link prediction [Lü and Zhou, 2011]....
[...]
929 citations
References
39,297 citations
"Link prediction in complex networks..." refers background in this paper
...The six real networks for testing are a protein–protein interaction network (PPI) [16], a co-authorship network of scientists who are themselves publishing on the topic of network science (NS) [53], an electrical power grid of the western US (Grid) [54], a network of the US political blogs (PB) [55], a router-level Internet collected by Rocketfuel Project (INT) [56], and a network of the US air transportation system (USAir) [57]....
[...]
...The reconstruction generally improves the estimates of clustering coefficient [54], modularity [133], assortativity [96,97], congestability,13 synchronizability14 and spreading threshold,15 indicating the validity of the approach....
[...]
35,161 citations
33,771 citations
"Link prediction in complex networks..." refers background in this paper
...Taking the power-law degree distribution as an instance, the well-known mechanisms include rich gets richer [45], good gets richer [140], optimal design [141], Hamiltonian dynamics [142], merging and regeneration [143], stability constraints [144], and so on....
[...]
...Since the groundbreaking work by Barabási and Albert [45], the evolving models all the time lie in the center of the complex network study....
[...]
...The mechanism of preferential attachment can be used to generate evolving scalefree networks,where the probability that a new link is connected to the node x is proportional to kx [45]....
[...]
19,398 citations
"Link prediction in complex networks..." refers methods in this paper
...|V|-flod cross-validation) will be applied in Section 6.2. Two standard metrics are used to quantify the accuracy of prediction algorithms: area under the receiver operating characteristic curve (AUC) [26] and Precision [27,28]. In principle, a link prediction algorithm provides an ordered list of all non-observed links (i.e., U− ET ) or equivalently gives each non-observed link, say (x,y) ∈ U− ET, a s...
[...]
18,415 citations