Community Detection Framework Research Articles

Incremental Community Detection on Large Complex Attributed Network

Community detection on network data is a fundamental task, and has many applications in industry. Network data in industry can be very large, with incomplete and complex attributes, and more importantly, growing. This calls for a community detection technique that is able to handle both attribute and topological information on large scale networks, and also is incremental. In this article, we propose inc-AGGMMR, an incremental community detection framework that is able to effectively address the challenges that come from scalability, mixed attributes, incomplete values, and evolving of the network. Through construction of augmented graph, we map attributes into the network by introducing attribute centers and belongingness edges. The communities are then detected by modularity maximization. During this process, we adjust the weights of belongingness edges to balance the contribution between attribute and topological information to the detection of communities. The weight adjustment mechanism enables incremental updates of community membership of all vertices. We evaluate inc-AGGMMR on five benchmark datasets against eight strong baselines. We also provide a case study to incrementally detect communities on a PayPal payment network which contains users with transactions. The results demonstrate inc-AGGMMR’s effectiveness and practicability.

Uncovering local community structure on line graph through degree centrality and expansion

Local community detection algorithms are an important type of overlapping community detection methods. Local community detection methods identify local community structure through searching seeds and expansion process. In this paper, we propose a novel local community detection method on line graph through degree centrality and expansion (LCDDCE). We firstly employ line graph model to transfer edges into nodes of a new graph. Secondly, we evaluate edges relationship through a novel node similarity method on line graph. Thirdly, we introduce local community detection framework to identify local node community structure of line graph, combined with degree centrality and PageRank algorithm. Finally, we transfer them back into original graph. The experimental results on three classical benchmarks show that our LCDDCE method achieves a higher performance on normalized mutual information metric with other typical methods.

SC-Com: Spotting Collusive Community in Opinion Spam Detection

In many cases, our decision-making process is closely related to online reviews. However, there have been threats of opinion spams by hired reviewers increasingly, which aim to mislead potential customers by hiding genuine consumers’ opinions. Opinion spams should be filed up collectively to falsify true information. Fortunately, we can spot the possibility to detect them from their collusiveness. In this paper, we propose SC-Com, an optimized collusive community detection framework. It constructs the graph of reviewers from the collusiveness of behavior and divides a graph by communities based on their mutual suspiciousness. After that, we extract community-based and temporal abnormality features which are critical to discriminate spammers from other genuine users. We show that our method detects collusive opinion spam reviewers effectively and precisely from their collective behavioral patterns. In the real-world dataset, our approach showed prominent performance while only considering primary data such as time and ratings.

A gene module identification algorithm\xa0and its applications\xa0to identify gene modules and key genes of hepatocellular carcinoma

To further improve the effect of gene modules identification, combining the Newman algorithm in community detection and K-means algorithm framework, a new method of gene module identification, GCNA-Kpca algorithm, was proposed. The core idea of the algorithm was to build a gene co-expression network (GCN) based on gene expression data firstly; Then the Newman algorithm was used to initially identify gene modules based on the topology of GCN, and the number of clusters and clustering centers were determined; Finally the number of clusters and clustering centers were input into the K-means algorithm framework, and the secondary clustering was performed based on the gene expression profile to obtain the final gene modules. The algorithm took into account the role of modularity in the clustering process, and could find the optimal membership module for each gene through multiple iterations. Experimental results showed that the algorithm proposed in this paper had the best performance in error rate, biological significance and CNN classification indicators (Precision, Recall and F-score). The gene module obtained by GCNA-Kpca was used for the task of key gene identification, and these key genes had the highest prognostic significance. Moreover, GCNA-Kpca algorithm was used to identify 10 key genes in hepatocellular carcinoma (HCC): CDC20, CCNB1, EIF4A3, H2AFX, NOP56, RFC4, NOP58, AURKA, PCNA, and FEN1. According to the validation, it was reasonable to speculate that these 10 key genes could be biomarkers for HCC. And NOP56 and NOP58 are key genes for HCC that we discovered for the first time.

A joint optimization framework for better community detection based on link prediction in social networks

Real-world network data can be incomplete (e.g., the social connections are partially observed) due to reasons such as graph sampling and privacy concerns. Consequently, communities detected based on such incomplete network information could be not as reliable as the ones identified based on the fully observed network. While existing studies first predict missing links and then detect communities, in this paper, a joint optimization framework, Communities detected on Predicted Edges, is proposed. Our goal to improve the quality of community detection through learning the probability of unseen links and the probability of community affiliation of nodes simultaneously. Link prediction and community detection are mutually reinforced to generate better results of both tasks. Experiments conducted on a number of well-known network data show that the proposed COPE stably outperforms several state-of-the-art community detection algorithms.

Community Detection via Local Learning Based on Generalized Metric With Neighboring Regularization

Community detection has long been a fundamental problem in network analysis. A great deal of previous research has regarded community detection as an optimization process, where a variety of internal quality metrics are typically treated as objective functions, such as modularity ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${Q}$ </tex-math></inline-formula> ) and weighted community clustering (WCC). However, purely optimizing a predefined quality metric probably results in an extreme unbalance in the scale of the detected communities, e.g., few giant communities with many very small communities. To reveal the true mesoscopic structure inside a big network under the suitable number of communities, we propose a novel community detection framework called LL-GMR, which is a local learning framework based on generalized metric with neighboring regularization. LL-GMR is qualified for both nonoverlapping and overlapping detection tasks. In LL-GMR, we propose a generalized representation and illustrate that it can be instantiated into 12 well-known internal quality metrics. When the generalized metric is used as an objective function, we encode node-level and community-level neighborhood information into two regularization terms to alleviate the dilemma of unbalanced communities. The experimental results show that our LL-GMR consistently outperforms other state-of-the-art community detection approaches in terms of discovering ground-truth communities in six real-life networks.

An adaptive population control framework for ACO-based community detection

Abstract The community structure is one of the most important features of complex networks and has wide research and application prospects. To find the community structure, many researchers currently focus on natural heuristic methods, where an extraordinary swarm intelligence algorithm (i.e., the ant colony algorithm) is widely adopted to detect the potential community structures. However, the computational cost of such an algorithm is so high that it restricts the property and range of application. In this paper, we present a novel adaptive population control framework for ACO-based community discovery approaches to overcome the mentioned shortcomings. Specifically, this framework dynamically controls the number of ants based on the slope of the modularity and iterations. Such a framework is adopted in two different algorithms and we make corresponding comparison between this one and traditional ACO-based algorithms in six classical real networks and five synthetic datasets. Experiments show that ant colony algorithms with our proposed framework have evidently reduced time complexity and maintained the quality of community structure simultaneously.

Identifying Communities in Dynamic Networks Using Information Dynamics.

Identifying communities in dynamic networks is essential for exploring the latent network structures, understanding network functions, predicting network evolution, and discovering abnormal network events. Many dynamic community detection methods have been proposed from different viewpoints. However, identifying the community structure in dynamic networks is very challenging due to the difficulty of parameter tuning, high time complexity and detection accuracy decreasing as time slices increase. In this paper, we present a dynamic community detection framework based on information dynamics and develop a dynamic community detection algorithm called DCDID (dynamic community detection based on information dynamics), which uses a batch processing technique to incrementally uncover communities in dynamic networks. DCDID employs the information dynamics model to simulate the exchange of information among nodes and aims to improve the efficiency of community detection by filtering out the unchanged subgraph. To illustrate the effectiveness of DCDID, we extensively test it on synthetic and real-world dynamic networks, and the results demonstrate that the DCDID algorithm is superior to the representative methods in relation to the quality of dynamic community detection.

A random effects stochastic block model for joint community detection in multiple networks with applications to neuroimaging

To analyze data from multisubject experiments in neuroimaging studies, we develop a modeling framework for joint community detection in a group of related networks that can be considered as a sample from a population of networks. The proposed random effects stochastic block model facilitates the study of group differences and subject-specific variations in the community structure. The model proposes a putative mean community structure, which is representative of the group or the population under consideration but is not the community structure of any individual component network. Instead, the community memberships of nodes vary in each component network with a transition matrix, thus modeling the variation in community structure across a group of subjects. To estimate the quantities of interest, we propose two methods: a variational EM algorithm and a model-free “two-step” method called Co-OSNTF which is based on nonnegative matrix factorization. We also develop a resampling-based hypothesis test for differences between community structure in two populations both at the whole network level and node level. The methodology is applied to the COBRE dataset, a publicly available fMRI dataset from multisubject experiments involving schizophrenia patients. Our methods reveal an overall putative community structure representative of the group as well as subject-specific variations within each of the two groups, healthy controls and schizophrenia patients. The model has good predictive ability for predicting community structure in subjects from the same population but outside the training sample. Using our network level hypothesis tests, we are able to ascertain statistically significant difference in community structure between the two groups, while our node level tests help determine the nodes that are driving the difference.

Graph-based data clustering via multiscale community detection

We present a graph-theoretical approach to data clustering, which combines the creation of a graph from the data with Markov Stability, a multiscale community detection framework. We show how the multiscale capabilities of the method allow the estimation of the number of clusters, as well as alleviating the sensitivity to the parameters in graph construction. We use both synthetic and benchmark real datasets to compare and evaluate several graph construction methods and clustering algorithms, and show that multiscale graph-based clustering achieves improved performance compared to popular clustering methods without the need to set externally the number of clusters.

Community detection in networks: a game-theoretic framework

Real-world networks are often cluttered and hard to organize. Recent studies show that most networks have the community structure, i.e., nodes with similar attributes form a certain community, which enables people to better understand the constitution of the networks and thus gain more insights into the complicated networks. Strategic nodes belonging to different communities interact with each other to decide mutual links in the networks. Hitherto, various community detection methods have been proposed in the literature, yet none of them takes the strategic interactions among nodes into consideration. Additionally, many real-world observations of networks are noisy and incomplete, i.e., with some missing links or fake links, due to either technology constraints or privacy regulations. In this work, a game-theoretic framework of community detection is established, where nodes interact and produce links with each other in a rational way based on mutual benefits, i.e., maximizing their own utility functions when forming a community. Given the proposed game-theoretic generative models for communities, we present a general community detection algorithm based on expectation maximization (EM). Simulations on synthetic networks and experiments on real-world networks demonstrate that the proposed detection method outperforms the state of the art.

Modularity Based Community Detection in Heterogeneous Networks

Heterogeneous networks are networks consisting of different types of nodes and multiple types of edges linking such nodes. While community detection has been extensively developed as a useful technique for analyzing networks that contain only one type of nodes, very few community detection techniques have been developed for heterogeneous networks. In this paper, we propose a modularity based community detection framework for heterogeneous networks. Unlike existing methods, the proposed approach has the flexibility to treat the number of communities as an unknown quantity. We describe a Louvain type maximization method for finding the community structure that maximizes the modularity function. Our simulation results show the advantages of the proposed method over existing methods. Moreover, the proposed modularity function is shown to be consistent under a heterogeneous stochastic blockmodel framework. Analyses of the DBLP four-area dataset and a MovieLens dataset demonstrate the usefulness of the proposed method.

A Unified Framework for Community Detection and Network Representation Learning

Network representation learning (NRL) aims to learn low-dimensional vectors for vertices in a network. Most existing NRL methods focus on learning representations from local context of vertices (such as their neighbors). Nevertheless, vertices in many complex networks also exhibit significant global patterns widely known as communities. It's intuitive that vertices in the same community tend to connect densely and share common attributes. These patterns are expected to improve NRL and benefit relevant evaluation tasks, such as link prediction and vertex classification. Inspired by the analogy between network representation learning and text modeling, we propose a unified NRL framework by introducing community information of vertices, named as Community-enhanced Network Representation Learning (CNRL). CNRL simultaneously detects community distribution of each vertex and learns embeddings of both vertices and communities. Moreover, the proposed community enhancement mechanism can be applied to various existing NRL models. In experiments, we evaluate our model on vertex classification, link prediction, and community detection using several real-world datasets. The results demonstrate that CNRL significantly and consistently outperforms other state-of-the-art methods while verifying our assumptions on the correlations between vertices and communities.

The evaluation and application of node influence in dynamic networks based on evolving communities

AbstractThe network connectivity in dynamic networks depends on a small number of highly mobile nodes. Identifying the influential nodes is one of the most engaging challenges for mobile applications, such as data offloading or worm propagation control. Reachability is an important metric to uncover node influence. Both TRGs (temporal reachability graphs) and CJEGs (critical journey evolving graphs) provide approaches to calculate reachability. Nevertheless, these approaches are only for epidemic scenarios. In practice, due to node privacy or limited battery life, message transmission is, to a certain extent, a probabilistic dissemination. Accordingly, reachability is difficult to exactly determine. As a structure of tight‐knit nodes, a community is born of a coarse‐grained estimation of reachability under a probabilistic propagation scenario. Based on an existing overlapping community detection framework, ie, AFOCS (an unsupervised machine learning algorithm), we propose an evolving overlapping community detection algorithm, ie, EFOCS, and further developed a metric, ie, OR_CEN, to estimate the reachability under the probabilistic propagation scenario. A content delivery experiment showed that OR_CEN accurately reveals the influence of nodes in dynamic networks. Based on OR_CEN, we also propose several target set selection algorithms and discuss their application in mobile data offloading. Analysis and simulation experiments indicated that CBS_OR, the target set selection algorithm based on OR_CEN centrality, has more advantages in scalability and distributed computing than CBS_AFOCS (an algorithm based on an aggregated AFOCS community), TRG_GREEDY (a greedy algorithm based on TRGs) and RANDOM (an algorithm based on a random selection strategy). Moreover, CBS_OR exhibited a significant better offloading effect than the algorithms mentioned earlier.

A framework of community detection based on individual labels in attribute networks

Community detection is an important problem for understanding the structure and function of complex networks and has attracted a lot of attention in recent decades. Most community detection algorithms only focus on the topology of networks. However, there is still much valuable information hidden in the networks, such as the attributes or content of the nodes and the useful prior information. Obviously, taking full advantage of these resources can improve the effectiveness of community detection. In this paper, we present a semi-supervised community detection framework named SCDAN (Semi-supervised Community Detection in Attribute Networks), in which a non-negative matrix factorization model is utilized to effectively integrate network topology, node attributes and individual labels simultaneously. The comparative experiments on real-world networks show that SCDAN significantly improves the performance of community detection and provides semantic interpretation of communities.

Constrained information flows in temporal networks reveal intermittent communities.

Many real-world networks represent dynamic systems with interactions that change over time, often in uncoordinated ways and at irregular intervals. For example, university students connect in intermittent groups that repeatedly form and dissolve based on multiple factors, including their lectures, interests, and friends. Such dynamic systems can be represented as multilayer networks where each layer represents a snapshot of the temporal network. In this representation, it is crucial that the links between layers accurately capture real dependencies between those layers. Often, however, these dependencies are unknown. Therefore, current methods connect layers based on simplistic assumptions that do not capture node-level layer dependencies. For example, connecting every node to itself in other layers with the same weight can wipe out dependencies between intermittent groups, making it difficult or even impossible to identify them. In this paper, we present a principled approach to estimating node-level layer dependencies based on the network structure within each layer. We implement our node-level coupling method in the community detection framework Infomap and demonstrate its performance compared to current methods on synthetic and real temporal networks. We show that our approach more effectively constrains information inside multilayer communities so that Infomap can better recover planted groups in multilayer benchmark networks that represent multiple modes with different groups and better identify intermittent communities in real temporal contact networks. These results suggest that node-level layer coupling can improve the modeling of information spreading in temporal networks and better capture intermittent community structure.

Fast detection of community structures using graph traversal in social networks

Finding community structures in social networks is considered to be a challenging task as many of the proposed algorithms are computationally expensive and does not scale well for large graphs. Most of the community detection algorithms proposed till date are unsuitable for applications that would require detection of communities in real-time, especially for massive networks. The Louvain method, which uses modularity maximization to detect clusters, is usually considered to be one of the fastest community detection algorithms even without any provable bound on its running time. We propose a novel graph traversal-based community detection framework, which not only runs faster than the Louvain method but also generates clusters of better quality for most of the benchmark datasets. We show that our algorithms run in O(|V | + |E|) time to create an initial cover before using modularity maximization to get the final cover. Keywords - community detection; Influenced Neighbor Score; brokers; community nodes; communities

A Probabilistic Framework for Structural Analysis and Community Detection in Directed Networks

There is growing interest in structural analysis of directed networks. Two major points that need to be addressed are: 1) a formal and precise definition of the graph clustering and community detection problem in directed networks and 2) algorithm design and evaluation of community detection algorithms in directed networks. Motivated by these, we develop a probabilistic framework for structural analysis and community detection in directed networks based on our previous work in undirected networks. By relaxing the assumption from symmetric bivariate distributions in our previous work to bivariate distributions that have the same marginal distributions in this paper, we can still formally define various notions for structural analysis in directed networks, including centrality, relative centrality, community, and modularity. We also extend three commonly used community detection algorithms in undirected networks to directed networks: the hierarchical agglomerative algorithm, the partitional algorithm, and the fast unfolding algorithm. These are made possible by two modularity preserving and sparsity preserving transformations. In conjunction with the probabilistic framework, we show these three algorithms converge in a finite number of steps. In particular, we show that the partitional algorithm is a linear time algorithm for large sparse graphs. Moreover, the outputs of the hierarchical agglomerative algorithm and the fast unfolding algorithm are guaranteed to be communities . These three algorithms can also be extended to general bivariate distributions with some minor modifications. We also conduct various experiments by using two sampling methods in directed networks: 1) PageRank and 2) random walks with self-loops and backward jumps.

A community detection algorithm based on multi-similarity method

Social network detection and identification constitute an important topic in the field of sociology. Previous graph similarity has focus on either the topological structure of graph or the feature value of vertex. In this work, a multi-similarity measure method for community is described. The approach devised by using multi-similarity properties based on vertex features, relationship density and topology structure, and therefore is can be formulated and extended to practical implementation. The framework of community detection combines K-means clustering, spectral clustering and modularity algorithm-making it an effective basis for the realization of a social network interpretation. With this scheme, three evaluation criteria are proposed for methodology determination. The experimental results show a better working performance of the recommended method than traditional algorithms via statistical analysis.

A uniform framework of projection and community detection for one-mode network in bipartite networks**Project supported by the National Natural Science Foundation of China (Grant Nos. 11505114 and 10975099), the Program for Professor of Special Appointment (Orientational Scholar) at Shanghai Institutions of Higher Learning (Grant Nos. QD02015016 and DUSST02), the Shanghai Project for Construction of Discipline Peaks, the Natural Science Foundation

Projection is a widely used method in bipartite networks. However, each projection has a specific application scenario and differs in the forms of mapping for bipartite networks. In this paper, inspired by the network-based information exchange dynamics, we propose a uniform framework of projection. Subsequently, an information exchange rate projection based on the nature of community structures of a network (named IERCP) is designed to detect community structures of bipartite networks. Results from the synthetic and real-world networks show that the IERCP algorithm has higher performance compared with the other projection methods. It suggests that the IERCP may extract more information hidden in bipartite networks and minimize information loss.