RolX: structural role extraction & mining in large graphs

By Keith W. Henderson, Brian Gallagher, Tina Eliassi-Rad, Hanghang Tong, Sugato Basu, Leman Akoglu, Danai Koutra, Christos Faloutsos, Lei Li. 2012

479 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Molecular Biology, Statistical and Nonlinear Physics, Graph Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Molecular Biology

In Statistical and Nonlinear Physics

In Graph Learning

Citations in the corpus

node2vec: Scalable Feature Learning for Networks (2016, background): “In contrast, under the structural equivalence hypothesis [10] nodes that have similar structural roles in networks should be embedded closely together ( e.g. , nodes u and s 6 in Figure 1 act as hubs of their corresponding communities).”
Representation Learning on Graphs: Methods and Applications (2017, background): “RolX ( B ) [29] is a baseline approach based upon hand-designed features, while struc2vec ( C ) and GraphWave ( D ) use different representation learning approaches.”
struc2vec: Learning Node Representations from Structural Identity (2017, methodology): “A recent approach to explicitly identify the role of nodes using just the network structure is RolX [7].”
Scalable Temporal Latent Space Inference for Link Prediction in Dynamic Social Networks (2014, methodology): “. 2.2 Inferring temporal latent space Existing approaches interpret the latent space model for nodes in networks either as roles of nodes or communities of nodes. Among the node-role based approaches [13, 22, 8], matrix factorization approaches are applied to user-feature matrix in each time step, t”
Detecting cohesive and 2-mode communities indirected and undirected networks (2014, background): “In contrast, the notion of structural equivalence suggests that nodes with similar connectivity patterns may be considered a community even if they do not link to each other [9, 17, 23].”
Learning Structural Node Embeddings via Diffusion Wavelets (2017, methodology): “A notable example of such approaches is RolX [11, 16], which aims to recover a soft-clustering of nodes into a predetermined number of K distinct roles using recursive feature extraction [17].”
Adversarial Graph Augmentation to Improve Graph Contrastive Learning (2021, unknown): “Experiments showed that these GNN models learn to over-emphasize node proximity [23] and may lose subtle but crucial structural information, thus failing in many tasks including node-role classiﬁcation [16,35,37,38] and graph classiﬁcation [17].”
Distance Encoding: Design Provably More Powerful Neural Networks for Graph Representation Learning (2020, background): “Then, the deﬁnition of structural representation provides a mathematical characterization the concept “structural roles” of nodes [3,5,6], where two far-away nodes could have the same structural roles (representations) as long as their ego-networks have the same structure.”
metapath2vec: Scalable Representation Learning for Heterogeneous Networks (2017, background): “5 RELATEDWORK Network representation learning can be traced back to the usage of latent factor models for network analysis and graph mining tasks [10, 34], such as the application of factorization models for recommendation systems [14, 16], node classication [32], relational mining [19], and role d”
Random walks and diffusion on networks (2016, unknown): “, “role similarity” [349, 416], “rich clubs” [417, 418], and approximately multipartite structure [419]).”
GCC: Graph Contrastive Coding for Graph Neural Network Pre-Training (2020, methodology): “and vto be the inner product of their representations. Finally, by sorting the above scores, we use HITS@10 (top-10 accuracy) to measure the performance of different methods. We compare GCC with RolX [18], Panther++ [66] and GraphWave [12]. We also provide random guess results for reference. Experim”
Graph based anomaly detection and description: a survey (2014, methodology): “…measures such as eigenvector [Bonacich and Lloyd, 2001], closeness [Noh and Rieger, 2004], and betweenness [Freeman, 1977] centralities, local clustering coefficient [Watts and Strogatz, 1998], radius [Kang et al., 2011c], degree assortativity, and most recently, roles [Henderson et al., 2012].”
Combining Label Propagation and Simple Models Out-performs Graph Neural Networks (2020, background): “There are numerous ways to get useful features from just the graph topology to augment the raw node features (Henderson et al., 2011; 2012; Hamilton et al., 2017b).”
Under review as a conference paper at ICLR 2019 MILE : A Multi-Level Framework for Scalable Graph Embedding (2018, unknown): “In recent years, graph embedding has attracted much interest due to its broad applicability for various tasks (Perozzi et al., 2014; Wang et al., 2016; Henderson et al., 2012).”