Real-world Social Network Datasets Research Articles

Graph Convolutional Networks with Markov Random Field Reasoning for Social Spammer Detection

The recent growth of social networking platforms also led to the emergence of social spammers, who overwhelm legitimate users with unwanted content. The existing social spammer detection methods can be characterized into two categories: features based ones and propagation-based ones. Features based methods mainly rely on matrix factorization using tweet text features, and regularization using social graphs is incorporated. However, these methods are fully supervised and can only utilize labeled part of social graphs, which fail to work in a real-world semi-supervised setting. The propagation-based methods primarily employ Markov Random Fields (MRFs) to capture human intuitions in user following relations, which cannot take advantages of rich text features. In this paper, we propose a novel social spammer detection model based on Graph Convolutional Networks (GCNs) that operate on directed social graphs by explicitly considering three types of neighbors. Furthermore, inspired by the propagation-based methods, we propose a MRF layer with refining effects to encapsulate these human insights in social relations, which can be formulated as a RNN through mean-field approximate inference, and stack on top of GCN layers to enable end-to-end training. We evaluate our proposed method on two real-world social network datasets, and the results demonstrate that our method outperforms the state-of-the-art approaches.

Community-diversified influence maximization in social networks

To meet the requirement of social influence analytics in various applications, the problem of influence maximization has been studied in recent years. The aim is to find a limited number of nodes (i.e., users) which can activate (i.e. influence) the maximum number of nodes in social networks. However, the community diversity of influenced users is largely ignored even though it has unique value in practice. For example, the higher community diversity reduces the risk of marketing campaigns as you should not put all your eggs in one basket; the diversity can also prolong the effect of a marketing campaign in the future promotion. Motivated by this observation, this paper investigates Community-diversified Influence Maximization (CDIM) problem to efficiently find k nodes such that, if a message is initiated and spread by the k nodes, the number as well as the community diversity of the activated nodes will be maximized at the end of propagation process. This work proposes a metric to measure the community-diversified influence and addresses a series of computational challenges. Two algorithms and an innovative CPSP-Tree index have been developed. This study also investigates the situation that community definition is not specified. The effectiveness and efficiency of the proposed solutions have been verified through extensive experimental studies on five real-world social network datasets.

Privacy Leakage via De-Anonymization and Aggregation in Heterogeneous Social Networks

Though representing a promising approach for personalization, targeting, and recommendation, aggregation of user profiles from multiple social networks will inevitably incur a serious privacy leakage issue. In this paper, we propose a Novel Heterogeneous De-anonymization Scheme (NHDS) aiming at de-anonymizing heterogeneous social networks. NHDS first leverages the network graph structure to significantly reduce the size of candidate set, then exploits user profile information to identify the correct mapping users with a high confidence. Performance evaluation on real-world social network datasets shows that NHDS significantly outperforms the prior schemes. Finally, we perform an empirical study on privacy leakage arising from cross-network aggregation based on four real-world social network datasets. Our findings show that 39.9 percent more information is disclosed through de-anonymization and the de-anonymized ratio is 84 percent. The detailed privacy leakage of user demographics and interests is also examined, which demonstrates the practicality of the identified privacy leakage issue.

Diversified top-[formula omitted] maximal clique detection in Social Internet of Things

Social Internet of Things (SIoT), an IoT where things are autonomously capable of establishing relationships with other smart objects related to humans, allows them to interact within a social structure based on relationships. Importantly, exploiting the social structures of smart objects in SIoT is important for supervision and management of various services. Diversified top-k maximal clique, as a novel social structure, can be used for anomaly detection, and smart community detection from SIoT. However, the scalability of the existing approaches for detecting diversified top-k maximal cliques is becoming a significant challenge faced in the big graph. To this end, this paper proposes a novel diversified top-k maximal clique detection approach based on formal concept analysis. Specifically, we firstly prove the existence of equivalence relation between maximal cliques and equiconcepts which are a class of special concepts where the extent and intent are the same. Based on this equivalence relation, an efficient and innovative approach based on formal concept analysis for identifying diversified top-k maximal cliques is then further presented. Finally, three real-world social network datasets are adopted in experiments for the validation of effectiveness of our approach in SIoT.

Targeted influence maximization under a multifactor-based information propagation model

Information propagation modeling and influence maximization are two important research problems in viral marketing. When marketing information is given, how can the seed nodes be efficiently identified to maximize the spread of the information through the network? To answer this question, we consider multiple factors in information propagation, such as information content, social influence and user authority, and propose a multifactor-based information propagation model (MFIP). Then, we utilize the first-order influence of the nodes to approximate their influence and propose an efficient heuristic algorithm named weighted degree decrease (WDD) to select the seed nodes under the MFIP model. Experimental evaluations with four real-world social network datasets demonstrate the effectiveness and efficiency of our algorithm.

Multi-Dimensional Recommendation Scheme for Social Networks Considering a User Relationship Strength Perspective

Developing a computational method based on user relationship strength for multi-dimensional recommendation is a significant challenge. The traditional recommendation methods have relatively low accuracy because they lack considering information from the perspective of user relationship strength into the recommendation algorithm. User relationship strength reflects the degree of closeness between two users, which can make the recommendation system more efficient between users in pairs. This paper proposes a multi-dimensional comprehensive recommendation method based on user relationship strength. We take three main factors into consideration, including the strength of user relationship, the similarity of entities, and the degree of user interest. First, we introduce a novel method to generate a user candidate set and an entity candidate set by calculating the relationship strength between two users and the similarity between two entities. Then, the algorithm will calculate the user interest degree of each user in the user candidate set to each entity in the entity candidate set, if the user interest degree is larger than or equal to a threshold, this particular entity will be recommended to this user. The performance of the proposed method was verified based on the real-world social network dataset and the e-commerce website dataset, and the experimental result suggests that this method can improve the recommendation accuracy.

Domain-adversarial Network Alignment

Network alignment is a critical task in a wide variety of fields. Many existing works leverage on representation learning to accomplish this task without eliminating domain representation bias induced by domain-dependent features, which yield inferior alignment performance. This paper proposes a unified deep architecture ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DANA</i> ) to obtain a domain-invariant representation for network alignment via an adversarial domain classifier. Specifically, we employ the graph convolutional networks to perform network embedding under the domain adversarial principle, given a small set of observed anchors. Then, the semi-supervised learning framework is optimized by maximizing a posterior probability distribution of observed anchors and the loss of a domain classifier simultaneously. We also develop a few variants of our model, such as, direction-aware network alignment, weight-sharing for directed networks and simplification of parameter space. Experiments on three real-world social network datasets demonstrate that our proposed approaches achieve state-of-the-art alignment results.

Diffusion model based on shared friends-aware independent cascade

Due to the swift growth of social network and the large variety of its application, numerous research effort aims at predict or explain the information diffusion phenomenon and how the user behaviour changes according to the effect of social pressure. Applications like viral marketing, rumour controlling, and individual behaviour analysis are some areas where these researches are applied. Many models proposed to handle the problem of information diffusion. Such models mostly require disseminated probabilities to be assigned for each link in diffusion network. In this paper we address the problem of predicting the information diffusion cascade. The proposed model named “Shared Friends-Aware IC-Based Diffusion Model (SFA-ICBDM)” can be seen as extension to the well known diffusion model, Independent Cascade model(IC). (SFA-ICBDM) model make use of structural feature of diffusion network specifically the common friend and similarity in terms of behaviour among users as a probability of influencing among users, then predict the cascade of information dissemination. The experiment has been conducted using real world social network dataset (Meme Tracker). In sum, the proposed model has been compared with baseline model IC and related works, where it shows improvement over baseline and related works. In addition, some promising conclusions have been obtained.

Multi-Winner Contests for Strategic Diffusion in Social Networks

Strategic diffusion encourages participants to take active roles in promoting stakeholders’ agendas by rewarding successful referrals. As social media continues to transform the way people communicate, strategic diffusion has become a powerful tool for stakeholders to influence people’s decisions or behaviors for desired objectives. Existing reward mechanisms for strategic diffusion are usually either vulnerable to falsename attacks or not individually rational for participants that have made successful referrals. Here, we introduce a novel multi-winner contests (MWC) mechanism for strategic diffusion in social networks. The MWC mechanism satisfies several desirable properties, including false-name-proofness, individual rationality, budget constraint, monotonicity, and subgraph constraint. Numerical experiments on four real-world social network datasets demonstrate that stakeholders can significantly boost participants’ aggregated efforts with proper design of competitions. Our work sheds light on how to design manipulation-resistant mechanisms with appropriate contests.

SpeCH: A scalable framework for data placement of data-intensive services in geo-distributed clouds

The advent of big data analytics and cloud computing technologies has resulted in wide-spread research on the data placement problem. Since data-intensive services require access to multiple datasets within each transaction, traditional schemes of uniformly partitioning the data into distributed nodes, as employed by many popular data stores like HDFS or Cassandra, may cause network congestion thereby affecting system throughput. In this article, we propose a scalable and unified framework for data-intensive service data placement into geographically distributed clouds. The proposed framework introduces a new paradigm for partitioning a set of data-items into geo-distributed clouds using Spectral Clustering on Hypergraphs, and is therefore called SpeCH. Scaling spectral methods to large workloads is challenging, since computing the spectra of the hypergraph laplacian is a computationally intensive task. SpeCH provides two solutions to tackle this problem: (1) an algorithm, called SpectralApprox, that leverages randomized techniques for obtaining low-rank approximations of the hypergraph matrix with bounded guarantees, thereby significantly improving the efficiency of spectral clustering while also providing high quality solutions in practice; (2) an algorithm, called SpectralDist, that exploits the highly parallel nature of the spectral clustering algorithm and uses Apache Spark to speed-up the process while retaining the same quality guarantees as the exact algorithm. Additionally, being distributed in nature, SpectralDist enables SpeCH to perform data placement on workloads that require resources beyond the capacity of a single machine. Experiments on a real-world trace-based online social network dataset show that the SpeCH is effective, efficient, and scalable. Empirically, SpectralApprox is comparable in efficacy on the evaluated metrics, while being up to 10 times faster in execution time when compared to state-of-the-art techniques. On the other hand, though SpectralApprox is 7–8 times faster when compared to SpectralDist, in terms of efficacy on the evaluated metrics the latter is up to 50% better.

Trajectory-Based User Encounter Prediction Over Wireless Sensor Networks

People or friends may encounter with each other offline, when they have a location proximity. With the rapid development of the wireless sensor network, smart city applications can leverage the sensed data of people’s mobility or trajectory to predict their future encounter opportunity and then arrange their offline activities (e.g., meeting, travel) accordingly. This paper studies the encounter prediction problem of mobile users by mining the similarity between their sensed mobile trajectories. We define the similarity of two mobile trajectories both temporally and spatially, and then propose two approaches, namely a probabilistic similarity maximization algorithm and a machine leaning based prediction algorithm, for addressing the encounter prediction problem. Results over a real-world social network dataset show that the proposed recurrent neural network based model can predict the encounter of two users precisely, and it outperforms the probabilistic algorithm and other algorithms, in terms of the precision and F1 score.

Retrieving Hidden Friends: A Collusion Privacy Attack Against Online Friend Search Engine

Online social networks (OSNs) are providing a variety of applications for human users to interact with families, friends, and even strangers. One such application, the friend search engine, allows the general public to query individual users’ friend lists and has been gaining popularity recently. However, without proper design, this application may mistakenly disclose users’ private relationship information. Our previous work has proposed a privacy preservation solution that can effectively boost OSNs’ sociability while protecting users’ friendship privacy against attacks launched by individual malicious requestors. In this paper, we propose an advanced collusion attack, where a victim user’s friendship privacy can be compromised through a series of carefully designed queries coordinately launched by multiple malicious requestors. The effect of the proposed collusion attack is validated through synthetic and real-world social network data sets. The in-depth research on the advanced collusion attacks will help us design a more robust and secure friend search engine on OSNs in the near future.

ABNE: An Attention-Based Network Embedding for User Alignment Across Social Networks

User alignment across social networks can facilitate more information/knowledge transferring across networks and thereby benefit several applications, including social link prediction, cross-domain recommendation, and information diffusion. Several works try to learn a common subspace for networks by preserving the structural proximities, such that different contribution weights of neighbors are ignored as users were always connected by unweighted edges. In this paper, we propose an attention-based network embedding model that exploits the social structures for user alignment. In particular, two main components are contained in our model framework: a masked graph attention mechanism which tries to learn the alignment task driven attention weights by the supervision of pre-aligned user pairs, and an embedding algorithm tries to learn a common vector space by explicitly modeling the weighted contribution probabilities between follower-ships and followee-ships. With the learned weights and embeddings transferring between these two components, we construct a unified model for user embedding and alignment. Stochastic gradient descent and negative sampling are adopted for efficient learning and scalability. The extensive experiments on real-world social network data sets demonstrate the effectiveness and robustness of the proposed model compared with several state-of-the-art methods.

Integrating Social Links into Wireless Networks: Modeling, Routing, Analysis, and Evaluation

Social connections among network nodes have been well investigated as an additional opportunity in network design (e.g., in routing strategies and trusted networking). This paper presents a paradigm shift that explores the design and performance analysis of combining social links jointly with communication links for message delivery in wireless networks. In a combined multi-layer social and communication network, communication links are based on conventional wireless technologies (e.g., WiFi, Bluetooth) and social links are overlaid over a communication infrastructure (e.g., cellular network) that provides an alternative way for data transmission. The goal is to characterize the performance analytically when routing is designed by combining social and communication links. A distance discretization technique is applied to model the reliability and delay of message delivery. The analytical foundation is developed to analyze the end-to-end delay and success probability under various effects of persistent transmission, potential error in distance estimation, and mobility. Systematic routing strategies that employ network inference are then designed to improve the performance in different aspects, such as delivery delay, delivery success probability, and energy-saving. A network emulation testbed is implemented with actual radios and real-world social network datasets to measure the performance of a heterogeneous network with social and communication links. The results in this paper show that the integration of social links in wireless network routing as a multi-layer design leads to substantial performance improvement for delay and reliability of message delivery.

Constrained Active Learning for Anchor Link Prediction Across Multiple Heterogeneous Social Networks.

Nowadays, people are usually involved in multiple heterogeneous social networks simultaneously. Discovering the anchor links between the accounts owned by the same users across different social networks is crucial for many important inter-network applications, e.g., cross-network link transfer and cross-network recommendation. Many different supervised models have been proposed to predict anchor links so far, but they are effective only when the labeled anchor links are abundant. However, in real scenarios, such a requirement can hardly be met and most anchor links are unlabeled, since manually labeling the inter-network anchor links is quite costly and tedious. To overcome such a problem and utilize the numerous unlabeled anchor links in model building, in this paper, we introduce the active learning based anchor link prediction problem. Different from the traditional active learning problems, due to the one-to-one constraint on anchor links, if an unlabeled anchor link is identified as positive (i.e., existing), all the other unlabeled anchor links incident to account u or account v will be negative (i.e., non-existing) automatically. Viewed in such a perspective, asking for the labels of potential positive anchor links in the unlabeled set will be rewarding in the active anchor link prediction problem. Various novel anchor link information gain measures are defined in this paper, based on which several constraint active anchor link prediction methods are introduced. Extensive experiments have been done on real-world social network datasets to compare the performance of these methods with state-of-art anchor link prediction methods. The experimental results show that the proposed Mean-entropy-based Constrained Active Learning (MC) method can outperform other methods with significant advantages.

Identifying the same person across two similar social networks in a unified way: Globally and locally

With the increasing popularity of online social networks, people today typically own several such accounts (e.g., Facebook, Twitter, and Flickr). In this scenario, it arises an interesting and challenging problem: how to identify the same person in different social networks, which is known as the network reconciliation problem. Prior work on this problem first assumes that the relationships between users are homogeneous, and then makes use of the local features (degree, common mapped neighbors) to achieve high precision. However, this assumption does not hold in reality since users usually have different tie strengths between each other. In this paper, we remodel the reconciliation problem by considering the users’ heterogeneous relationships and propose a unified framework called UniRank for incorporating the local features and global features together. Based on UniRank, we design an efficient two-stage network reconciliation algorithm. First, we design a global matching algorithm to explore more seeds with fast speed. Second, for each explored seed, we design a breadth-first strategy based local matching algorithm to match more seeds. Extensive simulations on real-world and synthetic social network datasets show that our algorithm significantly improves the state-of-art algorithm by up to 9X in terms of F1 score even under very rough conditions.

Incremental Subgraph Feature Selection for Graph Classification

Graph classification is an important tool for analyzing data with structure dependency, where subgraphs are often used as features for learning. In reality, the dimension of the subgraphs crucially depends on the threshold setting of the frequency support parameter, and the number may become extremely large. As a result, subgraphs may be incrementally discovered to form a feature stream and require the underlying graph classifier to effectively discover representative subgraph features from the subgraph feature stream. In this paper, we propose a primal-dual incremental subgraph feature selection algorithm (ISF) based on a max-margin graph classifier. The ISF algorithm constructs a sequence of solutions that are both primal and dual feasible. Each primal-dual pair shrinks the dual gap and renders a better solution for the optimal subgraph feature set. To avoid bias of ISF algorithm on short-pattern subgraph features, we present a new incremental subgraph join feature selection algorithm (ISJF) by forcing graph classifiers to join short-pattern subgraphs and generate long-pattern subgraph features. We evaluate the performance of the proposed models on both synthetic networks and real-world social network data sets. Experimental results demonstrate the effectiveness of the proposed methods.

Selecting discriminative features in social media data: An unsupervised approach

The usage of high-dimensional data complicates data processing in social network area. Accordingly, the researchers are motivated to propose some novel approaches to overcome this challenge. One of the best solutions is extracting the effective information from data pool and discarding the unnecessary ones. Feature selection is a known technique which aims to distinguish the discriminative features. Because of the unlabeled nature of datasets in social network, an unsupervised feature selection algorithm might be a good scenario. In addition to features, we try to confront the inherently linked users in social network datasets. This is because a stronger unsupervised feature selection technique is needed to ignore the independent and identically distributed assumption of data. Hence, by optimizing a novel objective function in this paper, the top-ranked features are extracted for further processing. This objective function incorporates both the inter-relationship of users in addition to their features. An efficient iterative algorithm is also designed to optimize the proposed objective function. We compare our method with two supervised and unsupervised evaluation criteria on real-world social network datasets. The experimental results demonstrate the effectiveness of our proposed approach.

Deep Community Detection

A deep community in a graph is a connected component that can only be seen after removal of nodes or edges from the rest of the graph. This paper formulates the problem of detecting deep communities as multi-stage node removal that maximizes a new centrality measure, called the local Fiedler vector centrality (LFVC), at each stage. The LFVC is associated with the sensitivity of algebraic connectivity to node or edge removals. We prove that a greedy node/edge removal strategy, based on successive maximization of LFVC, has bounded performance loss relative to the optimal, but intractable, combinatorial batch removal strategy. Under a stochastic block model framework, we show that the greedy LFVC strategy can extract deep communities with probability one as the number of observations becomes large. We apply the greedy LFVC strategy to real-world social network datasets. Compared with conventional community detection methods we demonstrate improved ability to identify important communities and key members in the network.

An incremental algorithm for updating betweenness centrality and k-betweenness centrality and its performance on realistic dynamic social network data

The increasing availability of dynamically changing digital data that can be used for extracting social networks over time has led to an upsurge of interest in the analysis of dynamic social networks. One key aspect of dynamic social network analysis is finding the central nodes in a network. However, dynamic calculation of centrality values for rapidly changing networks can be computationally expensive, with the result that data are frequently aggregated over many time periods and only intermittently analyzed for centrality measures. This paper presents an incremental betweenness centrality algorithm that efficiently updates betweenness centralities or k-betweenness centralities of nodes in dynamic social networks by avoiding re-computations through the efficient storage of information from earlier computations. In this paper, we evaluate the performance of the proposed algorithms for incremental betweenness centrality and k-betweenness centrality on both synthetic social network data sets and on several real-world social network data sets. The presented incremental algorithm can achieve substantial performance speedup (3–4 orders of magnitude faster for some data sets) when compared to the state of the art. And, incremental k-betweenness centrality, which is a good predictor of betweenness centrality, can be carried out on social network data sets with millions of nodes.