Online social networks have transformed global communication, enabling instant interaction but also accelerating the spread of rumors and misinformation that threaten public trust. While much research targets rumor detection, strategies for controlling rumor diffusion remain limited. This study systematically reviews 62 peer-reviewed papers focusing on control-oriented epidemiological (compartmental) models to analyze and mitigate rumor propagation across six major databases: Web of Science, IEEE Xplore, ProQuest, ScienceDirect, Engineering Village, and ACM Digital Library. Results show the Susceptible-Infected-Recovered (SIR) model dominates, with 87.10% of studies adopting deterministic approaches. Approximately 35% consider heterogeneous networks, and 75.81% rely on synthetic datasets, which restricts real-world validation. Common control measures include education & behavioral interventions, media & public communication, often optimized via Pontryagin's Maximum Principle. The review emphasizes the need for stronger empirical validation and adaptive modeling to enhance rumor management and ensure informationintegrity.

Social media platforms have experienced a rapid increase in automated accounts known as social bots, which are capable of spreading misinformation, spam, and malicious content. Detecting these bots is essential to maintain the reliability and security of online social networks. This paper proposes an intelligent social bot detection framework that utilizes graphbased learning and contrastive learning techniques to accurately identify automated accounts. The system constructs a social interaction graph where each node represents a user and edges represent interactions between users. A contrastive learning mechanism is used to learn meaningful representations of user behavior from multiple graph views. The proposed model analyzes structural and behavioral patterns to distinguish genuine users from bot accounts. Experimental evaluation performed on benchmark social media datasets demonstrates high accuracy and improved detection performance compared to traditional machine learning models. A user-friendly interface is also developed to visualize datasets, graph structures, and prediction results, allowing researchers and administrators to monitor suspicious activities effectively. The proposed framework provides a reliable and scalable solution for detecting social bots in large-scale social media networks.

Understanding how sentiment propagates in signed networks is crucial for uncovering mechanisms behind opinion polarization, trust formation, and information cocoons in digital communities. This paper investigates the generation of signed edges, representing positive or negative sentiments, in online social networks. We propose an analytical framework that models the dynamic growth of sentiment as a diffusion process. By introducing a walker on an infinite one-dimensional lattice, we derive a time-fractional diffusion equation that captures subdiffusive, normal diffusive, and superdiffusive behaviors. The model is empirically validated using two large-scale temporal signed networks: RedditHyperlinks and Bitcoin OTC. Our findings reveal that sentiment diffusion exhibits distinct regimes depending on the stage of network evolution, providing a foundation for further theoretical analysis and applications in signed social networks.

The exponential growth of visual content shared across online social networks has driven significant advancements in automated image analysis, fundamentally transforming how digital platforms process and monetize user data. However, this unchecked proliferation of high-dimensional visual data poses severe threats to user privacy, exposing sensitive attributes such as biometric identifiers, geographic locations, and underlying social associations. This paper presents a comprehensive investigation into privacy-aware visual computing, proposing a novel algorithmic framework designed to decouple sensitive identity representations from utility-preserving semantic features. By integrating adversarial representation learning with localized differential privacy mechanisms within the feature extraction pipeline, we establish a robust methodology for sanitizing images prior to cloud-based analysis. We conduct an exhaustive examination of the inherent trade-offs between visual utility, defined as the accuracy of downstream tasks such as object detection and scene classification, and privacy leakage, measured by the success rate of unauthorized facial recognition and attribute inference attacks. Our empirical evaluations, executed across multiple large-scale visual datasets, demonstrate that the proposed hybrid obfuscation architecture significantly diminishes privacy vulnerabilities while incurring a negligible degradation in primary task performance. The findings provide a scalable and mathematically rigorous foundation for deploying privacy-preserving computer vision models in distributed social networking environments.

Understanding the dynamics of online communities is crucial for comprehending modern social interactions and information dissemination. This research aims to understand how communities unfold and behave over time in the online environment of social media platforms by presenting a framework based on temporal fusion of information about text and network-type data. By employing text classification within identified communities, we uncover the underlying mechanisms that drive community formation and evolution in the online space. A dynamic social network analysis further reveals how real-world circumstances influence the development and interactions within these communities. Finally, we have identified fourteen key elements based on social science theories that encapsulate the insights expected from social structure and dynamics, and we have used the introduced methodology to evaluate how each key element enhances our understanding of social media dynamics, resulting in presenting our framework as a suited methodology for discourse fragmentation analysis. The framework is validated through a case study analyzing X (Twitter) data during major national circumstances in the United States in 2020. The discrimination discourse was found at the center of our analyses, and sexism, racism, xenophobia, ableism, homophobia, and religious intolerance are the fragments of the main discourse. Results show that the cycle of emergence and dissolution of the communities is fast and very representative of the discourse fragments. Real-world circumstances can impact the discourse fragments and their dominance, and comparing the number of distinct communities and their overlap, we reveal how social media can contribute to the formation of echo chambers and exacerbate societal polarization. The analyses extend beyond this scope, utilizing the introduced key elements related to opinion dynamics and structural insights to produce a comprehensive discourse fragmentation analysis. The framework's ability to identify and track discourse fragmentation provides critical insights for misinformation risk assessment, enabling early detection of false narrative communities and their evolutionpatterns.

The exponential growth of online social networking, the proliferation of fraudulent and bot-driven accounts has emerged as a critical threat to platform integrity. These accounts are commonly exploited to disseminate false information, manipulate user behavior, and engage in various deceptive practices. Addressing this challenge requires a robust and intelligent detection mechanism capable of adapting to increasingly sophisticated evasion tactics. This paper introduces the Social Media Fake Account Detection and Prevention System (SMFADPS), a multi-layered analytical framework that assesses account authenticity through the evaluation of multiple behavioral signals. The system examines profile completeness, content credibility, follower growth irregularities, and repeated content patterns to generate intermediate suspicion scores across four specialized detection modules. An ensemble-based weighted scoring mechanism consolidates these scores into a unified risk rating, which is subsequently used to categorize accounts into distinct threat levels. The system is developed using Python, FastAPI for RESTful service delivery, and a dual-database configuration comprising PostgreSQL and MongoDB. Evaluation conducted on a large-scale dataset confirms that the multi-signal approach yields substantially higher detection accuracy and operational efficiency compared to conventional single-indicator systems.

The proliferation in the use of Online Social Networks has revolutionized information sharing and consumption, leading to the development of advanced techniques such as link prediction, recommendation systems, community detection, node classification, and network representation learning. However, the availability and quality of real-world datasets for testing these algorithms pose challenges. Synthetic signed datasets generated through signed network reconstruction models offer alternatives for algorithm testing and experimentation. This survey presents an overview of state-of-the-art signed network reconstruction modeling techniques, evaluates their performance through rigorous experimental analysis, explores real-world applications, discusses challenges and open research problems, and guides future research efforts in the field. By consolidating knowledge and providing insights into existing models, this survey contributes to advancing the understanding and improvement of signed network reconstruction modeling. Various research papers discussed in this survey along with publicly available links to their codes are available at: https://github.com/Aikta-Arya/Signed-Network-Reconstruction-Modeling

Online social networks have become crucial platforms for marketing, leveraging their vast user bases and capacity for rapid information dissemination. Influence maximization (IM) problem plays a central role in applications such as viral marketing and information propagation. Nevertheless, IM remains challenging due to the difficulty in balancing accuracy and efficiency with existing methods. To overcome these limitations, a discrete hybrid optimizer called PDNS-DODE (PageRank-based Diffusion Neighborhood Search and Discrete Osprey Optimization with Differential Evolution) was proposed. The algorithm incorporates an adaptive three-hop diffusion neighborhood search (PDNS) strategy based on PageRank centrality, which dynamically adjusts the search scope to enable broader global exploration and targeted local optimization. A PageRank-descending initialization strategy is introduced to improve population diversity. Furthermore, the Osprey Optimization Algorithm (OOA) and Differential Evolution (DE) are discretized with revised individual representation and update mechanisms for solving the influence maximization problem. Extensive experiments on seven real-world networks under both Independent Cascade and Linear Threshold models demonstrate the robustness and superiority of PDNS-DODE. It consistently outperforms seven state-of-the-art baselines, achieving statistically significant improvements (Wilcoxon test) and the highest overall ranking (Friedman test). These advancements are attained while maintaining competitive computational efficiency.

In online social networks, cyber threats such as spam, phishing, and misinformation propagate through communication links, making security monitoring a critical challenge. Traditional approaches often lead to overburdening certain network connections, resulting in inefficient surveillance and increased vulnerability. To address this, we introduce the Fuzzy Regular Equitable Fair Domination Graph (FREFDG) and apply Equitable Fair Edge Domination (EFEDS) as a strategic model for balanced security monitoring. This framework ensures that security resources are distributed equitably across edges, thereby preventing overload and ensuring comprehensive threat detection. The theoretical foundations of EFEDS are rigorously established through propositions, theorems, and numerical illustrations, demonstrating its effectiveness in optimizing network surveillance. A decision-making model is formulated using graph-based analysis, enabling the systematic selection of critical edges requiring direct monitoring while leveraging indirect supervision for non-critical edges. This structured approach reduces computational complexity while enhancing network resilience. Additionally, we present a real-world application scenario, showcasing how EFEDS can be implemented in spam detection, phishing prevention, and misinformation filtering. The results demonstrate the efficiency of our framework in identifying high-risk edges, reducing redundant monitoring efforts, and improving threat mitigation strategies. By integrating fuzzy logic with equitable fair domination principles, our approach contributes to a more adaptive, scalable, and intelligent cybersecurity model, offering valuable insights for network security experts and researchers.

The article provides a review of mathematical models for social processes. Empirical research focuses mainly on measuring and analyzing network structures and data, while mathematical models focus more on understanding and predicting the mechanisms of social interactions. The paper contains search results with a brief description of the article selection process and their characteristics: 48 studies were presented in a detailed review, and 23 studies were included in this review based on neural network search. In addition, direct and inverse problems for the mean-field game model are formulated and numerically investigated in the case of stopping and stimulating the process of information dissemination in synthetic online social networks.

The multifaceted concept of body image (BI) refers to an individual's attitudes and impressions of their body. Negative BI is associated with a number of harmful health consequences, including obesity, eating disorders, and symptoms of sadness. The contemporary digital era, marked by the dominance of platforms, has brought about a considerable transformation in the landscape of BI issues. This study's goal is to compile and assess the connections between social media (SM) use, body weight, and BI in adult populations. This is a narrative review that comprehensively searches across multiple academic databases, such as PubMed, Medline, Scopus, Web of Science, and Google Scholar. Studies that used SM (online blogs, microblogs, content communities, or social networking sites) for engagement (e.g., sharing, commenting, liking) or image-related activities (e.g., viewing, posting, or engaging with images) with healthy adults (aged 18-70 years) of any body mass index (BMI kg/m2) met the inclusion criteria. Included were observational and experimental studies that examined habitual SM use. Only peer-reviewed works published in English between 2015 and 2025 met the search criteria. The currently available findings suggest that obese people are more dissatisfied with their bodies than people of normal weight, and obese women are more dissatisfied with their bodies than their peers of normal weight. Furthermore, experimental studies have demonstrated that immediate BI is adversely affected by acute exposure to idealized social media photographs. Policies should support specialized training that emphasizes a holistic approach to health and puts functionality and health above attractiveness. This training is crucial for dispelling weight-related stigmas and enabling healthcare providers to offer compassionate treatment that supports mental and physical health. Future research must concentrate on internalization and social pressure or reinforcement because these subjects have not gotten as much emphasis in prior studies. Such mechanism research could help better contextualize the role of recently introduced SM items.

Online social networks (OSNs) produce large volumes of user-generated data, enabling personalized services but also exposing users to significant privacy risks, a lack of transparency and frequent security breaches. Existing blockchain- and machine learning–based privacy-preservation methods struggle with high computational costs, limited scalability and weak malicious-node detection. To address these gaps, this work proposes a Blockchain-Driven Privacy Preservation Scheme with Progressive Graph Convolutional Networks (BPPS-SPD-PGCN) for secure and efficient protection of personal data in OSNs. The framework integrates Adaptive Two-Stage Unscented Kalman Filtering for data preprocessing, PGCN for malicious-node detection, ARPO for optimizing PGCN weights and Fair Proof-of-Reputation blockchain for secure access control. Two smart contracts (RG-SH and RG-ST) further enhance data confidentiality and storage integrity. Using the Epinions dataset, the proposed technique was evaluated through Accuracy, Precision, Recall, F1-score and Computational Time. The system achieved 99.04% Accuracy, 92.34% Precision, 99.14% Recall and 99.93% F1-score, outperforming PPB-OSN-GCN, HCS-PSC-SVM and BDI-ISPP-CNN. Overall, BPPS-SPD-PGCN provides a more robust, precise and secure privacy-preservation solution for OSNs, offering significant improvements over existing approaches.

Online discussions are often characterized by strong behavioral asymmetries: a relatively small fraction of users actively produces content, while the majority primarily consumes and redistributes it. Here we propose a community-detection framework for online social networks that exploits this asymmetry by first identifying and clustering a set of leading users, and then extending the resulting labels to the broader user base. We introduce two complementary strategies to cluster leaders, one based on their mutual interactions and the other on audience overlap, both relying on entropy-based filtering to separate signal from noise. We evaluate the framework on three major Italian political debates on Twitter/X, using public figures-identified through the pre-2022 verification system-as leaders, and known affiliations of political actors as ground truth labels. Compared with standard baselines, the proposed approach yields more coherent and interpretable communities aligned with political structures, with the two variants respectively recovering parties and coalitions. Activity-based criteria for selecting leaders produce qualitatively similar but consistently weaker results, particularly at the coalition level. Overall, our findings show that creating statistically validated networks of publicly recognized figures, whose off-platform roles constrain and stabilize their online behavior, provides a strong basis to identify discursive communities on social media. Although developed for Twitter/X, the approach is conceptually general, as it leverages structural asymmetries common to many online platforms.

In the era of pervasive online social networks (OSNs), the erosion of information privacy is occurring at an unprecedented rate. Empowering individuals with user-centric control over their private information is crucial to fostering public confidence in OSN services. Hence, the investigation into the personalized privacy configurations within the framework of differential privacy for OSNs, particularly for social relationships, is captivating. In this paper, we introduce a Collaborative Personalized Edge Differential Privacy model (CPEDP), ensuring personalized protection for sensitive social relationships while retaining the high utility of network features. Specifically, CPEDP allows each user to define a policy specification consisting of two complementary components: secret specifications at the edge level to identify sensitive relationships, and privacy specifications at the user level to determine personalized privacy parameters. These user-defined preferences are integrated through a collaborative privacy decision-making process that ensures consistent and interpretable privacy guarantees. Furthermore, we formalize the privacy primitive of CPEDP and develop a sampling-based mechanism to effectively implement the proposed model. Finally, comparative experiments on real-world datasets confirm that CPEDP achieves superior privacy-utility trade-offs, yielding more accurate estimates of key graph statistics through policy-driven personalization.

Online social networks have become essential platforms for communication and information sharing, yet their centralized architectures expose users to serious privacy, security, and integrity risks. Blockchain technology has emerged as a promising paradigm for addressing these challenges by enabling decentralization, transparency, immutability, and user-centric control over data. This survey aims to systematically examine blockchain-based solutions designed to enhance security and privacy in social networks. Following the PRISMA methodology, we review and analyze peer-reviewed studies published between 2016 and 2025, focusing on blockchain architectures, consensus mechanisms, smart contracts, and their integration with machine learning and deep learning techniques for malicious behavior detection. The survey categorizes existing blockchain-based social networks, compares centralized and decentralized models, and evaluates key trade-offs such as scalability versus security and privacy versus transparency. Additionally, regulatory and governance implications are discussed. The findings indicate that blockchain can significantly improve privacy, security, data integrity, and user empowerment in social networks; however, challenges related to scalability, usability, interoperability, and legal compliance remain. This survey provides a comprehensive overview for researchers and practitioners and identifies open research directions for the secure evolution of social networks.

Millet microgreens are novel functional foods which combine nutrient density and climatic resilience of millets with bioactive-rich profile of microgreens. Although they have the potential to enhance health, sustainability and food security, there is limited knowledge and consumption by the population. The paper was intended to evaluate knowledge, awareness, perception, and consumption behaviour of millet microgreens among the general population. It was conducted using a descriptive, online, self-administered questionnaire (Google Forms) and received more than 250 responses on different age, gender, and educational groups. Awareness levels, information sources, perceived health benefits, frequency of consumption, preferences, and barriers to use were used to analyze data using descriptive statistics. The general knowledge of millet microgreens was moderate, particularly among younger and more educated participants, but the level of nutritional knowledge and actual consumption was poor. The social media was the most common source of information, then it was family and peers, showing the impact of online platforms and social networks on food preferences. The majority of the respondents perceived millet microgreens to be healthier than regular vegetables and related them to better immunity and prevention of diseases, but said that they rarely or seldom ate them, which is a clear discrepancy between positive attitude and eating. The major challenges identified were the lack of availability in the market, perceived expensive price and the inability to know how to cook with millet microgreens. Positively, a significant proportion of the respondents showed interest in cultivating millet microgreens at their homes, which means that there is a high potential of uptake at the household level with the provision of specific nutrition education, practical advice, and access enhancement programs.

Abstract Racism remains a persistent societal issue, increasingly amplified by the structure and dynamics of online social networks. In this work, we propose a three-state compartmental model to study the spreading and suppression of racist content, drawing from epidemic-like dynamics and interaction-driven transitions. We analyze the model on fully connected (homogeneous mixing) networks using a set of coupled differential equations, and on Barabási–Albert scale-free and Watts–Strogatz small-world networks through agent-based simulations. The system exhibits three distinct stationary regimes: two racism-free absorbing states and one active phase with persistent racist content. We identify and characterize the phase transitions between these regimes, discuss the role of network topology, and highlight the emergence of absorbing states. Our findings illustrate how statistical physics tools can help uncover the macroscopic consequences of microscopic social interactions in digital environments. Graphical abstract

Various malicious activities performed by the social bots have brought a crisis of trust to the online social networks. In this paper, we propose a social bot detection method, named Accou2vec, based on community walk. First, in order to cut off the attacking edges between the human and bot accounts, the deep autoencoder-like non-negative matrix factorization community detection algorithm is leveraged to divide the social graph into multiple subgraphs. Then, we design the community walk rule that controls the intra-community walk and inter-community walk differently, considering both the number of nodes and edges in the community. Subsequently, the graph representation learning is used to learn the representation vector of each account. Finally, the representation vectors of labeled social bots and human accounts are used to train the classifier for social bots detection. Extensive experimental results on two real-world datasets show the superior performance of the proposed method over the state-of-the-art.

The development of online social networks is accompanied by intricate abnormal interaction phenomena severely impairing the ecosystem’s credibility. Current anomaly detection approaches find it challenging to balance accuracy and robustness when tackling dynamic structural changes, heterogeneous relationships, and lack of labeled data. To address these challenges, this paper proposes ST-MVAN, a Spatio-Temporal Multi-View Attention Network for unsupervised anomaly detection. The proposed framework integrates three core components: (1) in the spatial dimension, we construct heterogeneous relational subgraphs and design an improved Graph Convolutional Network (GCN) that incorporates edge attributes as additive bias and leverages sparse attention to filter structural noise; (2) for feature fusion, an Efficient Channel Attention (ECA) mechanism is introduced to adaptively assign importance weights to multi-view features; and (3) in the temporal dimension, a bidirectional GRU captures dynamic evolutionary dependencies. Finally, a joint Encoder-Decoder framework calculates anomaly scores based on reconstruction errors. Furthermore, we perform experiments on the Digg and Yelp datasets to validate that our method achieves an AUC improvement of up to 12.26% compared to baseline methods. These results demonstrate that ST-MVAN can effectively mitigate structural noise and enhance the security of dynamic social network environments.

Abstract In recent years, the relationship between the dimensions of online parental mediation (OPM) and minors’ use of screens and social networks has been studied. However, there remains a lack of consensus regarding the most effective strategies to promote adequate psychosocial adjustment, leading to the need to broaden the focus of OPM and analyze it from alternative perspectives. The objectives are: (1) to identify the profiles of OPM perceived by the children; (2) to relate these profiles to the time spent on devices and social networks. This study involved 4371 students from 32 schools in 11 Spanish regions aged 11 to 15 years ( M age =12.52, SD = 1.04). A validated OPM questionnaire with six dimensions was used: active mediation of internet use, active mediation of internet safety, child-initiated mediation, parental monitoring, technical controls, and restrictive mediation. In addition, ad hoc variables on the usage time of devices and social networks were examined. OPM was analyzed with latent profile analysis and chi-square comparisons with Bonferroni correction. The results revealed four mediation profiles: integral mediation (IM) (20.9%), proactive mediation (PM) (25.6%), technological mediation (TM) (26.2%), and minimal mediation (MM) (27.3%). The IM profile, which reflects a global and balanced approach to all the dimensions, was significantly associated with lower screen and social media usage times ( p ≤. 001), whereas MM was related to higher usage time ( p ≤. 001). Concerning the intermediate profiles, TM was generally related to less usage time than PM. The findings suggest that the amount of mediation is more relevant than the specific strategy, providing a more comprehensive view of OPM in today’s digital context.