From biological to social networks: Link prediction based on multi-way spectral clustering

Symeonidis, Panagiotis; Iakovidou, Nantia; Mantas, Nikolaos; Manolopoulos, Yannis

doi:10.1016/j.datak.2013.05.008

Cited by 55 publications

(33 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The AUC value can be explicated as the probability that a randomly chosen missing link is given a higher score than a randomly chosen nonexistent link. Through n independent comparisons, if there are n times the missing link having a higher score and n times they have the same score [2][32], then define AUC by Equation (8) as: In the proposed methods of this paper there are three parameters:…”

Section: Experimental Results and Comparisonmentioning

confidence: 99%

“…Moreover, if the value of eigenvalue equal to 0 this means that there is a connected components in a graph. Thus, spectral clustering is more elastic than k-means, in capturing the non-connected components of a graph [2].…”

Section: Spectral Clusteringmentioning

confidence: 99%

“…Notice that each element in the diagonal of matrix A is equal to zero to depict that a node is not connected to itself. In case of a large graph A can be characterized by high dimensionality [2].…”

Section: Preliminaries In Graphsmentioning

confidence: 99%

“…where D is the degree matrix and A is the adjacency matrix of graph G [2]. The goal of spectral clustering is to partition the data into k disjoint classes such that each x i belongs to one and only one class.…”

Section: Spectral Clusteringmentioning

confidence: 99%

See 3 more Smart Citations

Link Prediction in Social Networks based on Spectral Clustering using k-medoids and Landmark

Mahmoud¹,

Bakrawy²,

Ghali³

2017

IJCA

View full text Add to dashboard Cite

Social networks have become popular because they provide many services and advantages to its users like helping them to connect with new people, share content and share opinions with likeminded people. To conclude which new interactions between members are likely to occur in the near future? This problem defined as the link prediction problem. The traditional methods of link prediction are based on path features and graph features but few consider clustering information. So two link prediction methods based on spectral clustering using k-medoids and landmark are proposed. The first method uses k-medoids to cluster nodes of the graph based on eigenvectors that obtained from the normalized Laplacian matrix. While the second method selects a subset of data points as the landmarks and represents the original data points as the linear combinations of these landmarks. Experimental results demonstrate that our methods achieve high accuracy compared with LSk-means method (Link prediction using Spectral clustering by k-means). General TermsLink prediction, landmark based spectral clustering, k-medoids algorithm

show abstract

Section: Experimental Results and Comparisonmentioning

confidence: 99%

Section: Spectral Clusteringmentioning

confidence: 99%

Section: Preliminaries In Graphsmentioning

confidence: 99%

Section: Spectral Clusteringmentioning

confidence: 99%

See 2 more Smart Citations

Link Prediction in Social Networks based on Spectral Clustering using k-medoids and Landmark

Mahmoud¹,

Bakrawy²,

Ghali³

2017

IJCA

View full text Add to dashboard Cite

show abstract

“…For example, Symeonidis et al [22] did link prediction for biological and social networks based on multiway spectral clustering. Wang et al [23] and Krishna et al [24] predicted PPI interactions through matrix factorization-based methods.…”

Section: Introductionmentioning

confidence: 99%

Inference of protein-protein interaction networks from multiple heterogeneous data

Huang

Liao

2015

Proceedings of the 6th ACM Conference on Bioinformatics, Computational Biology and Health Informatics

View full text Add to dashboard Cite

Protein-protein interaction (PPI) prediction is a central task in achieving a better understanding of cellular and intracellular processes. Because high-throughput experimental methods are both expensive and time-consuming, and are also known of suffering from the problems of incompleteness and noise, many computational methods have been developed, with varied degrees of success. However, the inference of PPI network from multiple heterogeneous data sources remains a great challenge. In this work, we developed a novel method based on approximate Bayesian computation and modified differential evolution sampling (ABC-DEP) and regularized laplacian (RL) kernel. The method enables inference of PPI networks from topological properties and multiple heterogeneous features including gene expression and Pfam domain profiles, in forms of weighted kernels. The optimal weights are obtained by ABC-DEP, and the kernel fusion built based on optimal weights serves as input to RL to infer missing or new edges in the PPI network. Detailed comparisons with control methods have been made, and the results show that the accuracy of PPI prediction measured by AUC is increased by up to 23 %, as compared to a baseline without using optimal weights. The method can provide insights into the relations between PPIs and various feature kernels and demonstrates strong capability of predicting faraway interactions that cannot be well detected by traditional RL method.

show abstract

The Significant Effect of Overlapping Community Structures in Signed Social Networks

Shahriari

Liu

Klamma

2017

Lecture Notes in Social Networks

View full text Add to dashboard Cite

Many real-world networks can be modeled by networks of interacting agents. Analysis of these interactions can reveal fundamental properties from these networks. Estimating the amount of collaboration in a network corresponding to connections in a learning environment can reveal to what extent learners share their experience and knowledge with other learners. Alternatively, analyzing the network of interactions in an open source software project can manifest indicators showing the efficiency of collaborations. One central problem in such domains is the low cooperativity values of networks due to the low cooperativity values of their respective communities. So administrators should not only understand and predict the cooperativity of networks but also they need to evaluate their respective community structures. To approach this issue, in this paper, we address two domains of open source software projects and learning forums. As such, we calculate the amount of cooperativity in the corresponding networks and communities of these domains by applying several community detection algorithms. Moreover, we investigated the community properties and identified the significant properties for estimating the network and community cooperativity. Correspondingly, we identified to what extent various community detection algorithms affect the identification of significant properties and prediction of cooperativity. We also fabricated binary and regression prediction models using the community properties. Our results and constructed models can be used to infer cooperativity of community structures from their respective properties. When predicting high defective structures in networks, administrators can look for useful drives to increase the collaborations.

show abstract

From biological to social networks: Link prediction based on multi-way spectral clustering

Cited by 55 publications

References 42 publications

Link Prediction in Social Networks based on Spectral Clustering using k-medoids and Landmark

Link Prediction in Social Networks based on Spectral Clustering using k-medoids and Landmark

Inference of protein-protein interaction networks from multiple heterogeneous data

The Significant Effect of Overlapping Community Structures in Signed Social Networks

Contact Info

Product

Resources

About