Complex Networks & Their Applications XII

Simplicial complexes provide a useful framework of higher-order networks that model co-occurrences and interactions among more than two elements, and are also equipped with a mathematical foundation in algebraic topology. In this paper, we investigate the temporal growth processes of simplicial complexes derived from human activities and communication on social media from a perspective of burst vertices. First, we empirically show that most of new simplices contain burst vertices, while for each new simplex containing a burst vertex, its vertices other than the corresponding burst vertex do not necessarily co-occur with the burst vertex itself within the not so distant past. We thus examine the problem of finding which burst vertex is contained in a new simplex from the occurrence history of burst vertices. In particular, we focus on analyzing the influence of the occurrence events of burst vertices in terms of time-decays. To this end, we propose a probabilistic model incorporating a log-normal-like time-decay factor and give its learning method. Using real social media datasets, we demonstrate the significance of the proposed model in terms of prediction performance, and uncover the time-decay effects of burst vertices in the occurrence of new simplices by applying the proposed model.

In recent years, there has been a growing recognition that higher-order structures are important features in real-world networks. A particular class of structures that has gained prominence is known as a simplicial complex. Despite their application to complex processes such as social contagion and novel measures of centrality, not much is currently understood about the distributional properties of these complexes in communication networks. Furthermore, it is also an open question as to whether an established growth model, such as scale-free network growth with triad formation, is sophisticated enough to capture the distributional properties of simplicial complexes. In this paper, we use empirical data on five real-world communication networks to propose a functional form for the distributions of two important simplicial complex structures. We also show that, while the scale-free network growth model with triad formation captures the form of these distributions in networks evolved using the model, the best-fit parameters are significantly different between the real network and its simulated equivalent. An auxiliary contribution is an empirical profile of the two simplicial complexes in these five real-world networks.

The human brain, with its intricate network of neurons and synapses, remains one of the most complex systems to understand and model. The study presents a groundbreaking approach to understanding complex neural networks by introducing a dynamic fitting method for hybrid time-delayed and uncertain internally-coupled complex networks. Specifically, the research focuses on integrating a Neural Mass Model (NMM) called Jansen-Rit Model (JRM) with the Kuramoto model, by utilizing real human brain structural data from Diffusion Tensor Imaging (DTI), as well as functional data from Electroencephalography (EEG) and Magnetic Resonance Imaging (MRI), the study extends above two models into a more comprehensive brain-like model. This innovative multimodal model enables the simultaneous observation of frequency variations, synchronization states, and simulated electrophysiological activities, even in the presence of internal coupling and time delays. A parallel fast heuristics algorithm serves as the global optimization method, facilitating rapid convergence to a stable state that closely approximates real human brain dynamics. The findings offer a robust computational tool for neuroscience research, with the potential to simulate and understand a wide array of neurological conditions and cognitive states. This research not only advances our understanding of complex neural dynamics but also opens up exciting possibilities for future interdisciplinary studies by further refine or expand upon the current model.

This paper presents an adaptive network model in the context of joint action and social bonding. Exploration of mechanisms for mental and social network models are presented, specifically focusing on adaptation by bonding based on homophily and Hebbian learning during joint rhythmic action. The paper provides a comprehensive explanation of these concepts and their role in controlled adaptation within illustrative scenarios.

The research reported here analyses the relationship between group synchrony and group bonding through a novel adaptive computational dynamical system model. By simulating multimodal interactions within a group of four agents, the study uncovers patterns in group cohesion in the sense of emerging multimodal group synchrony and related group bonding. Findings include patterns for emerging group synchrony (logarithmical) and group bonding (logistic). The obtained insights offer an understanding of group interaction dynamics. Future research may consider larger groups and more variations of synchrony detection functions to widen the obtained findings.

In this paper, a first-order adaptive self-modeling network model is introduced to model information overload in the context of cyclical usage of smartphone apps. The model consists of interacting attention resources and emotional responses to both attention taxation and the app engagements. The model makes use of first-order reification to simulate the agent’s learning of the connections between app engagement and emotional responses, and strategic use of attention resources. Furthermore, external factors, such as context and influence of the environment to use the apps, are included to model the usage decision of the agent. Simulations in two scenarios illustrate that the model captures expected dynamics of the phenomenon.

The investigation of consumption behaviour on the level of every single person or some certain groups put forward some new tasks different from the behavioural analysis of the whole population. One of them is the problem of temporal peculiarities of consumer behaviour, for instance, how to find those, who react on some critical events faster than the others. It could be useful for identifying a focus-group which would show the tendency and help to make more accurate predictions for the rest of the population. A graph-based method of consumer’s behaviour analysis in the state space is developed in this research. The moments when the deviations from the usual behavioural trajectory occur are detected by incremental comparing the transition graph with its previous state. These moments collected for all customers help to separate the population by the delay time of their reaction to the critical situation. It’s also noticed that the velocity of the reaction is a personal feature of a customer, hence, this separation stays actual for different external events which cause the behavioural anomalies.

Although making mistakes is a crucial part of learning, it is still often being avoided in companies as it is considered as a shameful incident. This goes hand in hand with a mindset of a boss who dominantly believes that mistakes usually have negative consequences and therefore avoids them by only accepting simple tasks. Thus, there is no mechanism to learn from mistakes. Employees working for and being influenced by such a boss also strongly believe that mistakes usually have negative consequences but in addition they believe that the boss never makes mistakes, it is often believed that only those who never make mistakes can be bosses and hold power. That’s the problem, such kinds of bosses do not learn. So, on the one hand, we have bosses who select simple tasks to be always seen as perfect. Therefore, also they believe they should avoid mistakes. On the other hand, there exists a mindset of a boss who is not limited to simple tasks, he/she accepts more complex tasks and therefore in the end has better general performance by learning from mistakes. This then also affects the mindset and actions of employees in the same direction. This paper investigates the consequences of both attitudes for the organizations. It does so by computational analysis based on an adaptive dynamical systems modeling approach represented in a network format using the self-modeling network modeling principle.

In this paper, we investigate whether there is a standardized writing composition for articles in Wikipedia and, if so, what it entails. By employing a Neural Gas approximation to the topology of our dataset, we generate a graph that represents various prevalent textual compositions adopted by the texts in our dataset. Subsequently, we examine significantly attractive regions within our graph by tracking the evolution of articles over time. Our observations reveal the coexistence of different stable compositions and the emergence and disappearance of certain unstable compositions over time.

Social live streaming services (SLSS) and other consumer-generated media (CGM) offer gamification to attract people. Virtual gifts/tips such as Twitch’s “bits” are examples of this and construct interactive relationships between live streamers and viewers. However, their impact on user behavior and how the collection rates from the platforms to collect a portion of the tips affect user behavior and the platform’s gained rewards have not been sufficiently analyzed. This study focuses on the fact that CGM including SLSS, is a type of public goods game, and we propose a model that considers tipping systems with collection by platforms. Using our agent-based simulated environment, we demonstrate that the effect of tipping on user behavior depends considerably on the preference for psychological or monetary rewards and on network positions, such as the number of degrees. We also show that appropriate collection rates maximize the platform’s rewards. We believe that our results contribute to the design and operation of SLSS platforms.

This study examines how engagement-maximizing recommender systems influence the visibility of Members of Parliament’s tweets in timelines. Leveraging engagement predictive models and Twitter data, we evaluate various recommender systems. Our analysis reveals that prioritizing engagement decreases the ideological diversity of the audiences reached by Members of Parliament and increases the reach disparities between political groups. When evaluating the algorithmic amplification within the general population, engagement-based timelines confer greater advantages to mainstream right-wing parties compared to their left-wing counterparts. However, when considering users’ individual political leanings, engagement-based timelines amplify ideologically aligned content. We stress the need for audits accounting for user characteristics when assessing the distortions introduced by personalization algorithms and advocate addressing online platform regulations by directly evaluating the metrics platforms aim to optimize, beyond the mere algorithmic implementation.

Numerous disinformation campaigns are operating on social networks to influence public opinion. Detecting these campaigns primarily involves identifying coordinated communities. As disinformation campaigns can take place on several social networks at the same time, the detection must be cross-platform to get a proper picture of it. To encode the different types of coordination, a multi-layer network is built. We propose a scalable coordination detection algorithm, adapted from the Louvain algorithm and the Iterative Probabilistic Voting Consensus algorithm. This algorithm is applied to the previously built multi-layer network. Users from different social networks are then embedded in a common space to link communities with similar interests. This paper introduces an interpretable and modular framework used on three datasets to prove its effectiveness for coordination detection method and to illustrate it with real examples.

In our study, we investigate the persistence of misinformation in social networks, focusing on the longevity of discussions related to misinformation. We employ the CoVaxxy dataset, which encompasses COVID-19 vaccine-related tweets, and classify tweets as reliable/unreliable based on non-credible sources/accounts. We construct separate networks for retweets, replies, and mentions, applying centrality metrics (degree, betweenness, closeness) to assess tweet significance. Our objective is to determine how long tweets associated with non-credible sources remain active. Our findings reveal a noteworthy correlation: tweets with longer lifespans tend to be influential nodes within the network, while shorter-lived tweets have less impact. y shedding light on the longevity of misinformation within social networks, our research contributes to a better understanding of misinformation propagation dynamics. These insights can inform strategies to combat misinformation during public health crises like the COVID-19 pandemic.

Understanding information cascades in social networks is a critical research area with implications in various domains, such as viral marketing, opinion formation, and misinformation propagation. In information cascade prediction problem, one of the most important factors is the cascade structure of the social network, which can be described as a cascade graph, global graph, or an r-reachable graph. However, the majority of existing studies primarily focus on a singular type of relationship within the social network, relying on the homogeneous graph neural network. We introduce two novel approaches for heterogeneous social network cascading and analyze whether heterogeneous social networks have higher predictive accuracy than homogeneous networks, taking into account the potential differential effects of temporal sequences on the models. Further, our work highlights that the selection of edge types plays an important role in the accuracy of predicting information cascades within social networks.

While information about some environmental crises rapidly spreads following the initial event, other crises, such as the Flint Water Crisis (FWC), take years to garner national attention. Understanding the spread of information from local to national scales is important, as this spread is often necessary to receive recognition and resources. One way to understand these dynamics is to use information diffusion models, which have been used to investigate information spread during crises. Though there have been several studies on why the FWC took so long to reach national attention, such modeling techniques have not been used to gain additional insights into factors that may contribute to this delay. To address this gap, our study uses an independent cascade diffusion model to examine the information spreading dynamics of two environmental crises: the FWC and the train derailment in East Palestine, Ohio. Our results demonstrate how standard independent cascade models, despite adequately capturing a fast-spreading crisis, may not be sufficient to explain the dynamics of a delayed diffusion. Flint’s dynamics were only captured by manipulating the contagiousness parameter throughout the simulation, implying that social factors hindered its ability to fit the diffusion paradigm of unimpeded spread.

Online communication has overcome linguistic and cultural barriers, enabling global connection through social media platforms. However, linguistic variety introduced more challenges in tasks such as the detection of hate speech content. Although multiple NLP solutions were proposed using advanced machine learning techniques, data annotation scarcity is still a serious problem urging the need for employing semi-supervised approaches. This paper proposes an innovative solution—a multilingual Semi-Supervised model based on Generative Adversarial Networks (GAN) and mBERT models, namely SS-GAN-mBERT. We managed to detect hate speech in Indo-European languages (in English, German, and Hindi) using only 20% labeled data from the HASOC2019 dataset. Our approach excelled in multilingual, zero-shot cross-lingual, and monolingual paradigms, achieving, on average, a 9.23% F1 score boost and 5.75% accuracy increase over baseline mBERT model.

With our increasingly refined online browsing habits, the demand for high-grade recommendation systems has never been greater. Improvements constantly target general performance, evaluation, security, and explainability, but optimizing for serendipitous experiences is imperative since a serendipity-optimized recommender helps users discover unforeseen relevant content. Given that serendipity is a form of genuine unexpected experiences and recommenders are facilitators of user experiences, we aim at leveraging weak ties to explore their impact on serendipity. Weak links refer to social connections between individuals or groups that are not closely related or connected but can still provide valuable information and opportunities. On the other hand, the underlying social structure of recommender datasets can be misleading, rendering traditional network-based approaches ineffective. For that, we developed a network-inspired clustering mechanism to overcome this obstacle. This method elevates the system’s performance by optimizing models for unexpected content. By leveraging group weak ties, we aim to provide a novel perspective on the subject and suggest avenues for future research. Our study can also have practical implications for designing online platforms that enhance user experience by promoting unexpected discoveries.

We suggest interaction-CASCADE as a combined model by extending the CASCADE as a probabilistic high-level model to consider the underlying components’ failure interaction graph, which could be derived using detailed models. In interaction-CASCADE, the total incurred overload after each component failure is the same as the CASCADE; however, the overload transfers to the out neighbors of the failed component given by the interaction graph. We first assume that the component’s initial loads are independent of their out- and in-degrees in the interaction graph and show that even though the process’s dynamics depend on the interaction graphs’ structure, the critical load beyond which the probability of total failure is significant does not change. We then discuss that assigning the lighter loads to components with higher in-degrees can shift the minimum critical load to higher values. Simulation results for random Erdős-Rényi and power-law degree distributed are provided and discussed.

We provide a novel problem of analyzing geographical road networks from a perspective of identifying critical streets for vehicular evacuation. In vehicular evacuation behaviors during disasters and emergency situations, the shortest distance routes are not necessarily the best. Instead, routes that are easier to traverse can be more crucial, even if they involve detours. Furthermore, evacuation destinations need not be limited to conventional facilities or sites; wider and better maintained streets can also be suitable. Therefore, we focus on streets as basic units of road networks, and address the problem of finding critical streets in a geographical road network, considering a scenario in which people efficiently move from specified starting intersections located around their residences to designated goal streets, following the routes of easiest traversal. In this paper, we first model a road network as a vertex-weighted graph obtained from its topological representation, where vertices and edges represent streets and intersections between them, respectively. The weight of each vertex reflects its ease of traversal. Next, we extend a recently introduced edge-centrality measure, salience, for our problem, and propose a method of detecting critical streets based on the vertex-weighted graph of topological representation by incorporating the notion of damping factor into it. Using real-world road network obtained from OpenStreetMap, we experimentally reveal the characteristics of the proposed method by comparing it with several baselines.

The disruption of transportation systems caused by natural hazards in one region can have significant consequences on the distribution of agricultural products and their export. In various regions of the world, climate change is expected to increase the likelihood of multiple natural hazards, such as landslides or floods. Being able to model how perturbations to transportation networks affect critical export routes is an important step toward making the system more resilient. Here, we analyze how disruptions to the Brazilian soybeans transportation network would impact export economics. We show that the impact to the Brazilian market can be important, with most of the main routes showing an impact of more than 10% on costs. This in turn can have a significant impact on the worldwide markets. We also show that mitigation measures can and should be taken to adapt to the network weaknesses, especially in the face of climate change.

This study delves into the differences between incremental and optimized network design, with a focus on tree-shaped water distribution networks (WDNs). The study evaluates the cost overhead of incremental design under two distinct expansion models: random and gradual. Our findings reveal that while incremental design does incur a cost overhead, this overhead does not increase significantly as the network expands, especially under gradual expansion. We also evaluate the cost overhead for the two tree-shaped WDNs of a city in Cyprus. The paper underscores the need to consider the evolution of infrastructure networks, answering key questions about cost overhead, scalability, and design efficacy.

Using data collected from one-week periods in 2021 and 2022, both before and after billionaire Elon Musk’s acquisition of Twitter, we generated Twitter retweet networks to examine the connection between Musk and hate groups as designated by the US Southern Poverty Law Center (SPLC) in three separate hate ideologies: white nationalists / alt-right, anti-Semitics, and anti-LGBTQ. Utilizing the configuration model to generate random retweet networks, we successfully found a direct link between Twitter users who retweet Musk and users who retweet several SPLC-defined hate groups. Results show that Musk’s Tweets and general rhetoric have a potential appeal to hateful users on Twitter.

This study delves into the privacy risks associated with user interactions in complex networks such as those generated on social media platforms. In such networks, potentially sensitive information can be extracted and/or inferred from explicitly user-generated content and its (often uncontrolled) dissemination. Hence, this preliminary work first studies an unsupervised model generating a privacy risk score for a given user, which considers both sensitive information released directly by the user and content propagation in the complex network. In addition, a supervised model is studied, which identifies and incorporates features related to privacy risk. The results of both multi-class and binary privacy risk classification for both models are presented, using the Twitter platform as a scenario, and a publicly accessible purpose-built dataset.

Network visualization plays an important role in exploring information about network nodes and structure. Recently, a new network visualization method, namely the Uniform Placement of Alters on Spherical Surface (U-PASS), was proposed for ego-centric networks. In this work, we extend the U-PASS method for visualizing a general undirected unweighted network, which consists of the edges between nodes, network clusters, and some nodes that are not neighbors to the ego. We develop a new criterion that can quantify the degree of uniformity of the network with a target node as the ego. The performance comparison to several state-of-the-art methods show the our extended method performs better in terms of the uniformity of node scattering on the spherical surface.

The friendship paradox implies that, on average, a person will have fewer friends than their friends do. Prior work has shown how the friendship paradox can lead to perception biases regarding behaviors that correlate with the number of friends: for example, people tend to perceive their friends as being more socially engaged than they are. Here, we investigate the consequences of this type of social comparison in the conceptual setting of content creation (“sharing”) in an online social network. Suppose people compare the amount of feedback that their content receives to the amount of feedback that their friends’ content receives, and suppose they modify their sharing behavior as a result of that comparison. How does that impact overall sharing on the social network over time? We run simulations over model-generated synthetic networks, assuming initially uniform sharing and feedback rates. Thus, people’s initial modifications of their sharing behavior in response to social comparisons are entirely driven by the friendship paradox. These modifications induce inhomogeneities in sharing rates that can further alter perception biases. If people’s responses to social comparisons are monotonic (i.e., the larger the disparity, the larger the modification in sharing behavior), our simulations suggest that overall sharing in the network gradually declines. Meanwhile, convex responses can sustain or grow overall sharing in the network. We focus entirely on synthetic graphs in the present work and have not yet extended our simulations to real-world network topologies. Nevertheless, we do discuss practical implications, such as how interventions can be tailored to sustain long-term sharing, even in the presence of adverse social-comparison effects.

In this paper, we study the behavior of users on Online Social Networks in the context of Covid-19 vaccines in Italy. We identify two main polarized communities: Provax and Novax. We find that Novax users are more active, more clustered in the network, and share less reliable information compared to the Provax users. On average, Novax are more toxic than Provax. However, starting from June 2021, the Provax became more toxic than the Novax. We show that the change in trend is explained by the aggregation of some contagion effects and the change in the activity level within communities. In fact, we establish that Provax users who increase their intensity of activity after May 2021 are significantly more toxic than the other users, shifting the toxicity up within the Provax community. Our study suggests that users presenting a spiky activity pattern tend to be more toxic.

This paper describes the challenges posed by pattern-of-life variations when carrying out automated detection of abnormal events (change detection) in longitudinal (over-time) social network data sets using standard social network measures. In this paper we present a new scheme for substantially removing pattern-of-life variations from longitudinal social network measures. This new approach is based on a model in which pattern-of-life variations are modeled as time-dependent periodic multiplicative weights on the likelihood of initiating a new post in a social network. Unfortunately, analysis of real-world social network data reveals that the time-dependent weights change over time as well. Therefore, an approach for adaptively determining the time-dependent periodic multiplicative weights has been developed. A complete methodology for Adaptive Multiplicative Compensation for Pattern-of-Life variations is described and the methodology is tested on a suitable social media data set. The impact of pattern-of-life variations on the test over-time data set is reduced by up to a factor of 4X by the algorithm presented. The impact on the occurrences of false positive events (labeling a time point as a “change” when it is not) and the impact on the occurrences of false negative events (labeling a time point as “normal” when it really represented a change) clear in the test data set.

The Contextual Focal Structure Analysis (CFSA) model is a sophisticated approach enhancing the discovery and interpretability of focal structure spreaders on social networks, similar to the users’ dynamic interactions on Twitter. Leveraging the power of the multiplex networks approach, the CFSA model organizes data into multiple layers, allowing for a comprehensive examination of various user activities and their interests within social networks. The core of the CFSA model uses the users-users network layer to capture the complex interactions between users and obtain a deeper understanding of users’ engagements on the platform. The CFSA model incorporates hashtag co-occurrence networks as the second layer; it helps unveil the associations and relationships between hashtags mentioned on Twitter. To evaluate the effectiveness of the CFSA model, the study focused on the Cheng Ho disinformation narrative within the Indo-Pacific region. This analysis utilized a substantial dataset comprising over 64,519 tweets and 20,000 hashtags collected between January 2019 and October 2022. The findings revealed users’ activities and the supplementary contexts established through their engagement with different hashtags. These insights shed light on the intricate interplay between users, communities, and the content that shapes the discourse within the Indo-Pacific region. Impactful contextual focal structure sets emerged as key elements driving the conversation in the examined disinformation narrative within the dataset. The CFSA model exposes significant patterns of popular hashtags such as “#SouthChinaSea,” “#NavyPartnerships,” and “#United_States”. Part of these hashtags were linked to accounts disseminating information concerning oil and gas exploration and drilling operations, mainly undertaken by the NATO alliances and China.

Various users with diverse spiritual moods and morals react differently to online content. On some users, highly toxic submissions cannot have a noticeable impact, while on another group, even not severely toxic content may provoke them to stop their interactive participation in social media. The same is true toward the intervention of platform holders, in the sense that one punishing moderation may lead a user to respect the community rules to a certain point, while the same action may stimulate the violent and offensive reaction of another user. In that regard, the moderation interventions of the platforms should follow a fine-level nature and be based on user behavior. It should also try to protect the communities the user cares about the most as far as the user, in turn, respects the content policies. The aim of the current study is to classify users into various behavioral groups, which can potentially provide the chance to adopt more efficient moderative measures to protect the community and give the user the feeling that he really deserves the moderative intervention that has experienced. Thus, the behavior of the core users of an already-banned controversial subreddit was taken into consideration, and a machine learning-based classification strategy was imposed on their activity level and their submission toxicity scores. Results have revealed interesting behavioral differences between users of the Reddit social media toward the taken moderations and have indicated the necessity for adopting find-level measures for protecting the platform, as well as the different behavioral groups.

Federated social networks have become an appealing choice as alternatives to mainstream centralized platforms. In the current global context, where the user’s activity on various social networks is monitored, influenced and manipulated, alternative platforms that offer the possibility of owning and controlling one’s own data are of great importance. Mastodon stands out among decentralized alternatives in the fediverse.In this study, we conduct a time-based dynamics analysis of Mastodon instances within a specific period. Our results show a growth pattern of instances in terms of accounts in certain periods of time, and due to social events, reinforcing our assumption of it being already trusted as a decentralized platform. Our work holds significance in the wider context of studying and understanding the adoption and evolution of decentralized platforms as ethical alternatives to Big Tech platforms.

This paper introduces the Better Hide Communities (BHC) benchmark dataset, purposefully crafted for gauging the efficacy of current and prospective community deception algorithms. BHC facilitates the evaluation of algorithmic performance in identifying the best set of updates to apply to a network to hide a target community from community detection algorithms. We believe that BHC will help in advancing the development of community deception algorithms and in promoting a deeper understanding of algorithmic capabilities in applying deceptive practices within communities.

Influential spreaders are used to maximize or control the spreading dynamics in a network. It acts as a maximizer in the case of information dissemination and a controller to control the epidemic spreading. In the literature, researchers are mostly focused on finding the best spreader from an undirected network. Indeed, the edge’s direction of a spreading process in the network has immense significance while estimating the influential spreaders. This paper presents a novel method, i.e., the “crossbred method” to identify the best spreaders for a directed network. The proposed method considers the spreading properties of the directed network. It takes account of two popular parameters of a spreading process, i.e., node’s out-degree and spreading’s reachability of an originator node. We have verified the spreading performance of the proposed method with the Directed Susceptible-Infected-Recovered (SIR) spreading epidemic model on six real networks. The outcome of the investigation demonstrates that the proposed method achieved significant improvement in terms of spreading dynamics over the existing methods of directed networks such as out-degree centrality, betweenness centrality, pagerank centrality, eigenvector centrality, cluster-rank centrality, outgoing closeness centrality, hybrid centrality.

Amidst the growth of harmful content on social media platforms, encompassing abusive language, disrespect, and hate speech, efforts to tackle this issue persist. However, effectively preventing the impact of such content on individuals and communities remains a challenging endeavor. In this paper, we present a study using Reddit data, where we employ a tree structure to visually and comprehensively examine the impact of toxic content on communities. By applying various machine learning algorithms, we classify the toxicity of each leaf node based on its parent and grandparent nodes, as well as the overall tree’s average toxicity. Our methodology can help policymakers detect early warning signs of toxicity and redirect potentially harmful comments to less toxic directions. Our research provides a comprehensive analysis of toxicity on social media platforms, allowing for a better understanding of differences and similarities across platforms, and a deeper exploration of the impact of toxic content on individual communities. Our findings provide valuable perspectives on the prevalence and consequences of toxic content on social media platforms, and our approach can be used in future studies to provide a more nuanced understanding of this complex issue.

The Uyghurs are a Turkic ethnic group that descends from the general region of Central and East Asia. In recent years, the Chinese government has been accused of violating the human rights of Uyghur Muslims by imprisoning them in camps and subjecting them to forced labor, forced abortion, and other harmful treatments. Aside from Western news outlets, much information about the Uyghur issue is available on blogs and social media. The information found in these blogs provides an alternative lens to analyze the Uyghur discourse from nontraditional media outlets. Existing studies have analyzed different contents of blogs, such as the authors and comments. This research aims to employ a systematic approach to analyze information spread through hyperlinks from Uyghur-related blog posts in the Indo-Pacific region. We analyzed 318 blog posts and 5,598 hyperlinks published from January 2019 to September 2022 to achieve this. The analysis included topic modeling and toxicity analysis to identify the main topics in each blog post and their toxicity trends. The topics generated from the topic modeling were mapped onto a network of hyperlinks to develop a topic-induced network. This network helped examine, analyze, and visualize the network of blog hyperlinks that further revealed influential blog conduits raising awareness about the oppression of the Uyghur community.

We investigate a method to asses brain synchronization in individuals who fulfill a cooperation task. Our input is a couple of signals from functional Near-Infrared Spectroscopy Data Acquisition and Pre-processing technology that is used to capture the brain activity of an individual by measuring the oxyhemoglobin (HbO) level. Then, we use the visibility graph approach to map each HbO signal into a network. We estimate the signal synchronization by studying a global measure, related to eigenvalues of Laplacian matrix, in each constructed visibility graph. We consider the autonomous evolution of one isolated node to be a Rössler function. Then, the synchronization of signals can be characterized by a little number of parameters that could be employed to classify the sources of signal. Unlike prior research in this area, our aim is to examine the circumstances in which synchronization occurs in various individuals and within different hemispheres of the prefrontal cortexes of the same individual. Experimental results show that the conditions for synchronization vary in different individuals, and they are different even for the distinct prefrontal cortical hemispheres of the same individual.

This article investigates synchronization analysis and verification for complex networked systems with nonlinear coupling. Based on general Lyapunov functions beyond the quadratic form, a less conservative synchronization criterion is proposed for the nonlinear networked systems under the directed topology. Then, the synchronization problem for polynomial networks can be converted into a sum-of-squares programming problem, which falls within the convex programming framework, yielding polynomial Lyapunov functions efficiently to realize the automatic synchronization verification in polynomial time. Finally, the effectiveness of the theoretical results is demonstrated by a numerical example, in which our proposed method can guarantee to achieve synchronization by using a smaller lower bound of coupling strength.

This article addresses the problem of resilient consensus for multi-agent networks. Resilience is used here to distinguish disruptive agents from compliant agents which follow a given control law. We present an algorithm enabling efficient and resilient network consensus based on an inversion of the social dynamics of the Deffuant model with emotions. This is achieved through the exploitation of a dynamic tolerance linked to extremism and clustering, whereby agents filter out extreme non-standard opinions driving them away from consensus. This method is not dependent on prior knowledge of either the network topology or the number of disruptive agents, making it suitable for real-world applications where this information is typically unavailable.

A social interaction (so-called higher-order event/interaction) can be regarded as the activation of the hyperlink among the corresponding individuals. Social interactions can be, thus, represented as higher-order temporal networks, that record the higher-order events occurring at each time step over time. The prediction of higher-order interactions is usually overlooked in traditional temporal network prediction methods, where a higher-order interaction is regarded as a set of pairwise interactions. We propose a memory-based model that predicts the higher-order temporal network (or events) one step ahead, based on the network observed in the past and a baseline utilizing pairwise temporal network prediction method. In eight real-world networks, we find that our model consistently outperforms the baseline. Importantly, our model reveals how past interactions of the target hyperlink and different types of hyperlinks that overlap with the target hyperlinks contribute to the prediction of the activation of the target link in the future.