Influence Propagation Research Articles

Graph embedding based ant colony optimization for negative influence propagation suppression under cost constraints

• The idea of swarm intelligence is introduced, and the positive feedback mechanism of ant colony is used to simulate the spread of negative influence. • Graph embedding technology is used to reconstruct the relationship between nodes in the network. • This method can select cost-effective suppression nodes with limited cost, and can effectively prevent the spread of the node’s negative influence in different communities. • This research can play an important role in the fields of epidemic prevention and control, rumor spreading and so on. In the information spreading mechanism of social networks, the influence propagation of information sources often has different effects on different users. How to effectively suppress the negative effects is particularly important. In the case of unknown network propagation principle, this paper introduces the idea of swarm intelligence, which utilizes the positive feedback mechanism of ant colony to simulate the propagation of negative influence, and finds a set of high-value and low-cost suppression nodes. On this basis, the graph embedding technique is used to obtain the new relationships between nodes in the network, and the new relationships between the nodes are used as heuristic information for the ant colony algorithm. Experiments show that our algorithm can not only find the set of inhibitory nodes with limited cost, but also effectively limit the spread of negative influence in the network compared with other algorithms. The research of this paper can not only enrich the theoretical research results of influence maximization, but also play an important role in the analysis of network topology, as well as in the fields of epidemic prevention and control, rumor propagation and so on.

Research on the Influence Maximization Problem in Social Networks Based on the Multi-Functional Complex Networks Model

Most of the existing influence maximization problem in social networks only focus on single relationship social networks, that is, there is only one relationship in social networks. However, in reality, there are often many relationships among users of social networks, and these relationships jointly affect the propagation of network information and its final scope of influence. Based on the classical linear threshold model and combined with various relationships between network nodes, in this paper MRSN-LT propagation model is proposed to model the influence propagation process between nodes in multiple relationships social networks. Then, MRSN-RRset algorithm based on reverse reachable set is proposed to solve the problem of low computational performance caused by greedy algorithm in the research process of traditional influence maximization. Finally, the experimental results on real data sets show that the proposed method has better influence propagation scope and greater computational performance improvement.

Learning Graph Structures With Transformer for Multivariate Time-Series Anomaly Detection in IoT

Many real-world IoT systems, which include a variety of internet-connected sensory devices, produce substantial amounts of multivariate time series data. Meanwhile, vital IoT infrastructures like smart power grids and water distribution networks are frequently targeted by cyber-attacks, making anomaly detection an important study topic. Modeling such relatedness is, nevertheless, unavoidable for any efficient and effective anomaly detection system, given the intricate topological and nonlinear connections that are originally unknown among sensors. Furthermore, detecting anomalies in multivariate time series is difficult due to their temporal dependency and stochasticity. This paper presented GTA, a new framework for multivariate time series anomaly detection that involves automatically learning a graph structure, graph convolution, and modeling temporal dependency using a Transformer-based architecture. The connection learning policy, which is based on the Gumbel-softmax sampling approach to learn bi-directed links among sensors directly, is at the heart of learning graph structure. To describe the anomaly information flow between network nodes, we introduced a new graph convolution called Influence Propagation convolution. In addition, to tackle the quadratic complexity barrier, we suggested a multi-branch attention mechanism to replace the original multi-head self-attention method. Extensive experiments on four publicly available anomaly detection benchmarks further demonstrate the superiority of our approach over alternative state-of-the-arts. Codes are available at https://github.com/ZEKAICHEN/GTA.

Continuous Influence-Based Community Partition for Social Networks

Community partition is of great importance in social networks because of the rapid increasing network scale, data and applications. We consider the community partition problem under LT model in social networks, which is a combinatorial optimization problem that divides the social network to disjoint $m$ communities. Our goal is to maximize the sum of influence propagation through maximizing it within each community. As the influence propagation function of community partition problem is supermodular under LT model, we use the method of Lov{$\acute{a}$}sz Extension to relax the target influence function and transfer our goal to maximize the relaxed function over a matroid polytope. Next, we propose a continuous greedy algorithm using the properties of the relaxed function to solve our problem, which needs to be discretized in concrete implementation. Then, random rounding technique is used to convert the fractional solution to integer solution. We present a theoretical analysis with $1-1/e$ approximation ratio for the proposed algorithms. Extensive experiments are conducted to evaluate the performance of the proposed continuous greedy algorithms on real-world online social networks datasets and the results demonstrate that continuous community partition method can improve influence spread and accuracy of the community partition effectively.

A New Strategy in Boosting Information Spread.

Finding a seed set to propagate more information within a specific budget is defined as the influence maximization (IM) problem. The traditional IM model contains two cardinal aspects: (i) the influence propagation model and (ii) effective/efficient seed-seeking algorithms. However, most of models only consider one kind of node (i.e., influential nodes), ignoring the role of other nodes (e.g., boosting nodes) in the spreading process, which are irrational. Specifically, in the real-world propagation scenario, the boosting nodes always improve the spread of influence from the initial activated seeds, which is an efficient and cost-economic measure. In this paper, we consider the realistic budgeted influence maximization (RBIM) problem, which contains two kind of nodes to improve the diffusion of influence. Facing the newly modified objective function, we propose a novel B-degree discount algorithm to solve it. The novel B-degree discount algorithm which adopts the cost-economic boosting nodes to retweet the influence from the predecessor nodes can greatly reduce the cost, and performs better than other state-of-the-art algorithms in both effect and efficiency on RBIM problem solving.

Finding Core-Periphery Structures With Node Influences

Detecting core-periphery structures is one of the outstanding issues in complex network analysis. Various algorithms can identify core nodes and periphery nodes. Recent advances found that many networks from real-world data can be better modeled with multiple pairs of core-periphery nodes. In this study, we propose to use an influence propagation process to detect multiple pairs of core-periphery nodes. In this framework, we assume each node can emit a certain amount of influence and propagate it through the network. Then we identify nodes with large influences as core nodes, and we utilize a maximum influence chain to construct a node-pairing network to determine core-periphery pairs. This approach can take node interactions into consideration and can reduce noises in finding pairs. Experiments on randomly generated networks and real-world networks confirm the efficiency and accuracy of our algorithm.

Propagation dynamics for an age-structured population model in time-space periodic habitat

How do environmental heterogeneity influence propagation dynamics of the age-structured invasive species? We investigate this problem by considering a yearly generation invasive species in time-space periodic habitat. Starting from an age-structured population growth law, we formulate a reaction-diffusion model with time-space periodic dispersal, mortality and recruitment. Thanks to the fundamental solution for linear part of the model, we reduce to study the dynamics of a time-space periodic semiflow which is defined by the solution map. By the recent developed dynamical theory in Fang et al. (J Funct Anal 272:4222-4262, 2017), we obtained the spreading speed and its coincidence with the minimal wave speed of time-space periodic traveling waves, as well as the variational characterization of spreading speed in terms of a principal eigenvalue problem. Such results are also proved back to the reaction-diffusion model.

Chinese named-entity recognition via self-attention mechanism and position-aware influence propagation embedding

Chinese Named Entity Recognition (NER) has received extensive research attention in recent years. However, Chinese texts lack delimiters to divide the boundaries of words, and some existing approaches cannot capture the long-distance interdependent features. In this paper, we propose a novel end-to-end model for Chinese NER. A new global word boundary detection approach is designed to capture the semantic dependency via a self-attention mechanism to represent character embedding by assigning compatible weights for each character in a sentence. To improve the representation ability of Chinese named-entity boundaries, we introduce position-aware influence propagation with the Gaussian kernel for each character, which combines convergence propagation and radiation propagation. Convergence propagation mainly measures the influence of surrounding characters on the target character. The purpose of radiation propagation is to measure the range of influence of the target character on surrounding characters. The proposed method has been evaluated and shown to offer strong performance in two Chinese NER datasets: MSRA and PFR.

Influence Minimization With Node Surveillance in Online Social Networks

Diffusion dynamics is the transfer of information from one node to another. Information diffusion has two goals: maximize or minimize information spread over the network. Attractive information such as innovations, awareness campaigns, branding, and advertising help people positively. However, awful information such as rumors, malicious viruses, pornography, and revenge disturb people. The negative information contributes to chaos; therefore, it must be blocked and inhibited from further diffusion. We are motivated to study the problem, namely <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Influence Minimization</i> . The new information alters the energy level or entropy associated with a node. Entropy quantifies the influence propagation rate across the network. This article proposes two reduction policies that reduce repulsive information’s influence through entropy. We validate the proposed system by considering the user response and surveillance on real-time networks.

Retracted] Research on Maximizing Influence of Blockchain Social Network Based on BCLT Model

In the blockchain social network, the traditional influence maximization algorithm has the problem of insufficient accuracy of the influence spread. To solve the above problem, a BCLT model including the characteristics of the blockchain is established based on the linear threshold model. The BC‐RIS algorithm is proposed based on the reverse reachable set. The BC‐RIS algorithm's influence spread and running time and the traditional algorithm is compared using the real blockchain social network data set. The experimental results show that the BC‐RIS algorithm can obtain a larger influence spread range, which is more in line with the influence propagation law of the blockchain social network.

Influence Maximization Based on Backward Reasoning in Online Social Networks

Along with the rapid development of information technology, online social networks have become more and more popular, which has greatly changed the way of information diffusion. Influence maximization is one of the hot research issues in online social network analysis. It refers to mining the most influential top-K nodes from an online social network to maximize the final propagation of influence in the network. The existing studies have shown that the greedy algorithms can obtain a highly accurate result, but its calculation is time-consuming. Although heuristic algorithms can improve efficiency, it is at the expense of accuracy. To balance the contradiction between calculation accuracy and efficiency, we propose a new framework based on backward reasoning called Influence Maximization Based on Backward Reasoning. This new framework uses the maximum influence area in the network to reversely infer the most likely seed nodes, which is based on maximum likelihood estimation. The scheme we adopted demonstrates four strengths. First, it achieves a balance between the accuracy of the result and efficiency. Second, it defines the influence cardinality of the node based on the information diffusion process and the network topology structure, which guarantees the accuracy of the algorithm. Third, the calculation method based on message-passing greatly reduces the computational complexity. More importantly, we applied the proposed framework to different types of real online social network datasets and conducted a series of experiments with different specifications and settings to verify the advantages of the algorithm. The results of the experiments are very promising.

Influence Maximization with Latency Requirements on Social Networks

Targeted marketing strategies are of significant interest in the smartapp economy. Typically, one seeks to identify individuals to strategically target in a social network so that the network is influenced at a minimal cost. In many practical settings, the effects of direct influence predominate, leading to the positive influence dominating set with partial payments (PIDS-PP) problem that we discuss in this paper. The PIDS-PP problem is NP-complete because it generalizes the dominating set problem. We discuss several mixed integer programming formulations for the PIDS-PP problem. First, we describe two compact formulations on the payment space. We then develop a stronger compact extended formulation. We show that when the underlying graph is a tree, this compact extended formulation provides integral solutions for the node selection variables. In conjunction, we describe a polynomial-time dynamic programming algorithm for the PIDS-PP problem on trees. We project the compact extended formulation onto the payment space, providing an equivalently strong formulation that has exponentially many constraints. We present a polynomial time algorithm to solve the associated separation problem. Our computational experience on a test bed of 100 real-world graph instances (with up to approximately 465,000 nodes and 835,000 edges) demonstrates the efficacy of our strongest payment space formulation. It finds solutions that are on average 0.4% from optimality and solves 80 of the 100 instances to optimality. Summary of Contribution: The study of influence propagation is important in a number of applications including marketing, epidemiology, and healthcare. Typically, in these problems, one seeks to identify individuals to strategically target in a social network so that the entire network is influenced at a minimal cost. With the ease of tracking consumers in the smartapp economy, the scope and nature of these problems have become larger. Consequently, there is considerable interest across multiple research communities in computationally solving large-scale influence maximization problems, which thus represent significant opportunities for the development of operations research–based methods and analysis in this interface. This paper introduces the positive influence dominating set with partial payments (PIDS-PP) problem, an influence maximization problem where the effects of direct influence predominate, and it is possible to make partial payments to nodes that are not targeted. The paper focuses on model development to solve large-scale PIDS-PP problems. To this end, starting from an initial base optimization model, it uses several operations research model strengthening techniques to develop two equivalent models that have strong computational performance (and can be theoretically shown to be the best model for trees). Computational experiments on a test bed of 100 real-world graph instances (with up to approximately 465,000 nodes and 835,000 edges) attest to the efficacy of the best model, which finds solutions that are on average 0.4% from optimality and solves 80 of the 100 instances to optimality.

Subsidence control method by inversely-inclined slicing and upward mining for ultra-thick steep seams

Ultra-thick steep coal seam mining will inevitably lead to the increase of greater and violent ground subsidence and deformation. A subsidence control method by inversely-inclined slicing and upward mining is proposed in this paper. By this method, the sequence of collapse of overlying strata and the direction of propagation of strata movement are changed, the extent of roof-side deformation thereby is lessened, and boundary angle of roof-side subsidence is reduced by 5°–10°. The mechanism of this mining method for control of strata movement has been evidenced by numerical simulation and experiments with similarity materials. A subsidence prediction model based on the variation of mining influence propagation angle can be used to evaluate the surface movement and deformation of the mining method. The application of the method in No.3 Mine in Yaojie mining area has yielded the expected result.

Multi-perspective social recommendation method with graph representation learning

Social recommender systems (SRS) aim to study how social relations influence users’ choices and how to use them for better learning users embeddings. However, the diversity of social relationships, which is instructive to the propagation of social influence, has been rarely explored. In this paper, we propose a graph convolutional network based representation learning method, namely multi-perspective social recommendation (MPSR), to construct hierarchical user preferences and assign friends’ influences with different levels of trust from varying perspectives. We further utilize the attributes of items to partition and excavate users’ explicit preferences and employ complementary perspective modeling to learn implicit preferences of users. To measure the trust degree of friends from different perspectives, the statistical information of users’ historical behavior is utilized to construct multi-perspective social networks. Experimental results on two public datasets of Yelp and Ciao demonstrate that the MPSR significantly outperforms the state-of-the-art methods. Further detailed analysis verifies the importance of mining explicit characteristics of users and the necessity for diverse social relationships, which show the rationality and effectiveness of the proposed model. The source Python code will be available upon request.

High Influencing Pattern Discovery over Time Series Data

A spatial co-location pattern denotes a subset of spatial features whose instances frequently appear nearby. High influence co-location pattern mining is used to find co-location patterns with high influence in specific aspects. Studies of such pattern mining usually rely on spatial distance for measuring nearness between instances, a method that cannot be applied to an influence propagation process concluded from epidemic dispersal scenarios. To discover meaningful patterns by using fruitful results in this field, we extend existing approaches and propose a mining framework. We first defined a new concept of proximity to depict semantic nearness between instances of distinct features, thus applying a star-shaped materialized model to mine influencing patterns. Then, we designed attribute descriptors to perceive attributes of instances and edges from time series data, and we calculated the attribute weights via an analytic hierarchy process, thereby computing the influence between instances and the influence of features in influencing patterns. Next, we constructed influencing metrics and set a threshold to discover high influencing patterns. Since the metrics do not satisfy the downward closure property, we propose two improved algorithms to boost efficiency. Extensive experiments conducted on real and synthetic datasets verified the effectiveness, efficiency, and scalability of our method.

SPEX: A Generic Framework for Enhancing Neural Social Recommendation

Social Recommender Systems (SRS) have attracted considerable attention since its accompanying service, social networks, helps increase user satisfaction and provides auxiliary information to improve recommendations. However, most existing SRS focus on social influence and ignore another essential social phenomenon, i.e., social homophily. Social homophily, which is the premise of social influence, indicates that people tend to build social relations with similar people and form influence propagation paths. In this article, we propose a generic framework Social PathExplorer (SPEX) to enhance neural SRS. SPEX treats the neural recommendation model as a black box and improves the quality of recommendations by modeling the social recommendation task, the formation of social homophily, and their mutual effect in the manner of multi-task learning. We design a Graph Neural Network based component for influence propagation path prediction to help SPEX capture the rich information conveyed by the formation of social homophily. We further propose an uncertainty based task balancing method to set appropriate task weights for the recommendation task and the path prediction task during the joint optimization. Extensive experiments have validated that SPEX can be easily plugged into various state-of-the-art neural recommendation models and help improve their performance. The source code of our work is available at: https://github.com/XMUDM/SPEX.

Competitive influence maximization considering inactive nodes and community homophily

Influence maximization has attracted much attention as a way to maximize influence spread through a social network. Most previous works focus on a non-competitive environment. But in the real world such as E-commerce, there are often multiple competitors at the same time to find influential users in order to advertise their similar products. So competitive influence maximization is introduced and some solutions are proposed. However, present studies have two limitations. First, they ignore the community homogeneity problem, which means that, information can spread easily among the individuals in the community, but it is hard to be diffused to other communities. Second, those latent users that are in inactive state are often considered to have no effects on other nodes. So most researches only consider those active nodes in terms of influence propagation. But in fact inactive nodes do have effects on the information propagation. In order to address above problems, in this paper we propose a competitive influence maximization method considering inactive nodes and community homophily. We first propose a new propagation model considering inactive nodes to simulate the information dissemination in a competitive environment, in which the inactive nodes are considered to have a weaker influence compared to the active ones. In order to break down the barriers of information dissemination between different communities, we propose a two-phase seed node selecting algorithm which has the minimum node number and maximum influence range. Our experiments on real world datasets verify the feasibility and accuracy of our proposed model and algorithm.

Node deletion-based algorithm for blocking maximizing on negative influence from uncertain sources

The spreading of negative influence, such as epidemic, rumor, false information and computer virus, may lead to serious consequences in social networks. The issue of negative influence blocking maximization arouses intense interest of the researchers. However, in the real world social network environment, the exact source of negative influence is usually unknown. In most cases, we only know the distribution of negative seeds, which is the probability for each node to be a negative seed. In this work, we investigate the problem of maximizing the blocking on negative influence from uncertain sources. We propose the competitive influence linear threshold propagation model (CI-LTPM) for the problem. Based on the IC-LTPM model, we define the problem of uncertain negative source influence blocking maximization (UNS-IBM). We use the propagation tree in the live-edge (LE) sub-graph for estimating the influence propagation. An algorithm is proposed to calculate the blocking increments of the positive seeds based on the propagation tree in the LE sub-graph. We observed that the blocking effect of the positive seeds is the reduction on the negative influence after the positive seeds and their related edges being deleted from the LE sub-graph. Based on such observation, we propose a node deletion-based algorithm NDB (node-deletion-blocking) for solving the UNS-IBM problem. Our experiment results show that NDB can block more negative influence in less computational time than other methods.

Time-sensitive Positive Influence Maximization in signed social networks

Influence maximization (IM) in social networks has attracted a wide variety of studies motivated by applications like spread of ideas or innovations in a network and viral marketing of products. How to integrate the factors affecting influence diffusion into influence maximization is an key research direction. Timeliness of influence diffusion and the polarity of influence itself are two important factors, which have been used to extend IM problem by researchers. However, all existing studies only focused on one of two factors individually, this will lead time delay of influence propagation or inaccurate influence estimation. Aiming at this drawback, we propose the Time-sensitive Positive Influence Maximization (TP-IM) problem considering two factors simultaneously, to select the seed node set achieving maximum positive influence spread within the specified time limit. Furthermore, we construct the Heat Diffusion-based Polarity Influence Diffusion (HDPID) model and an improved k-Step Greedy seed node selection algorithm, to solve the TP-IM problem. Experimental results on three signed social network datasets, Epinions, Slashdot, and Wikipedia demonstrate that our method outperforms state-of-the-art methods in terms of positive influence spread with time limit.

Launcher nodes for detecting efficient influencers in social networks

Influence propagation in social networks is a subject of growing interest. A relevant issue in those networks involves the identification of key influencers. These players have an important role on viral marketing strategies and message propagation, including political propaganda and fake news. In effect, an important way to fight malicious usage on social networks is to understand their properties, their structure and the way messages propagate.This paper proposes a new index for analyzing message propagation in social networks, based on the network topological nature and the influential power of the message. The new index characterizes the strength of each node as a launcher of the message, dividing the nodes into launchers and non-launchers. This division is most evident when the viral power of the message is high. Together with other known metrics, launcher individuals can assist to select efficient influencers in a social network. For instance, instead of choosing a strong member according to its degree in the network (number of followers), we may previously select those belonging to the launchers group and then look for the lowest degree members contained therein. These members are probably cheaper (on financial incentives) but still guarantying almost the same influence effectiveness as the largest degree members.We discuss this index using a number of real-world social networks available in known datasets repositories.