Modern enterprises increasingly leverage multicloud architectures, introducing complex security and compliance challenges. While Cloud-Native Application Protection Platforms (CNAPPs) offer unified solutions, most are cloud-hosted and require exporting sensitive configuration data, posing significant privacy and data sovereignty risks. This paper introduces Sentinel, a novel private-by-design CNAPP agent that embodies a new architectural pattern we term the Locally-Orchestrated, Privacy-Preserving Agent (LOPPA). Sentinel employs a modular, on-device execution model orchestrated by a fine-tuned DeepSeek language model, enabling automated compliance and threat analysis without exposing data to third-party services. This is particularly critical for high-value environments like the Intel Developer Cloud, where sensitive intellectual property is managed. We detail Sentinel’s innovative architecture, its privacy-preserving fine-tuning methodology, and the LOPPA model that maintains continuous compliance while eliminating data transfer. Our evaluation demonstrates that Sentinel achieves detection accuracy comparable to leading cloud-hosted alternatives (within a 2-3% F1 score) with acceptable performance trade-offs, establishing that secure, privacy-preserving AI agents are a viable and superior alternative for organizations with stringent data sovereignty requirements. This work advances privacy-preserving security automation through its novel combination of on-device LLM orchestration, modular compliance engines, and local execution guarantees.
While prior research has investigated the impact of human influencers (HIs), the escalation of virtual influencers (VIs) draws attention to how these personas influence their followers’ psychological and behavioral responses. The present study explores the impact of social networks on society and human behavior. Specifically, this study uncovers how engaging with human versus virtual social media influencers (SMIs) affects social comparison, fear of missing out (FOMO), phubbing behavior, social withdrawal, and mental health wellbeing among their followers. Data collected from 303 SMI were collected and analyzed using regression analysis in SPSS. Results demonstrate that human influencers have stronger negative effects on followers’ emotional, psychological, and mental well-being. Results contribute to theory and practice.
This paper proposes a machine learning-based firewall framework that enhances intrusion detection by applying a union of two metaheuristic algorithms—Bat Algorithm (BA) and Dragonfly Algorithm (DA)—for feature selection. The UNSWNB15 dataset is used, where the union of features selected by both algorithms improves model accuracy while reducing dimensionality. Two classifiers, Naïve Bayes (NB) and Random Forest (RF), are employed to evaluate the performance of the selected features. The Random Search technique is used to optimize hyperparameters. Experimental results show that the RF classifier achieved perfect scores across all evaluation metrics, while NB yielded near-perfect results. These findings confirm the effectiveness of combining BA and DA to improve the precision and efficiency of attack detection. The proposed model offers a promising solution for real-world cybersecurity applications where accurate and fast decision-making is essential.
As social networking sites (SNS) become central to information consumption, they increasingly shape how historical narratives are constructed, circulated, and remembered. This paper investigates how algorithmic curation, engagement bias, misinformation, memory distortion, and digital memory loss collectively influence historical memory in digital environments. Drawing from memory studies, media theory, and communication research, it proposes the Historical Narrative Formation Framework to conceptualize the recursive and interactive mechanisms through which digital platforms affect public recollection. Unlike existing studies that treat these dynamics in isolation, the framework maps their interdependencies, showing how emotionally charged, user-driven content interacts with algorithmic systems to reinforce certain narratives while erasing or distorting others. A conceptual case study of 9/11 reinterpretations on TikTok illustrates how platform dynamics reshape younger generations’ historical understanding through engagement-optimized content, conspiratorial framing, and selective forgetting. The paper concludes by highlighting the implications of platform governance, user agency, and algorithmic design in the longterm construction of collective memory, calling for greater critical literacy, archival safeguards, and attention to marginalized voices in digital historiography.
Community detection serves as a key component of network science, essential for revealing and examining structures within complex networks. This study offers a thorough examination that combines theoretical concepts with detailed practical evaluation. At the heart of our investigation are two interrelated methodologies: the Geodesic Distance Metric (GDM), which serves as a structure-aware, label-independent evaluation tool; and the Metric Backbone, a parameter-free sparsification approach that maintains essential community structures. We present and assess a scalable pipeline for community detection that utilizes Metric Backbone sparsification in conjunction with community detection algorithms, with outcomes evaluated through various well-known quality metrics. Our systematic empirical evaluation across seven diverse real-world datasets shows that Metric Backbone sparsification leads to significant edge reduction-ranging from $14 \%$ in sparse citation networks to $71 \%$ in dense social networks-while preserving community detection quality. Interestingly, the High School dataset maintains F1 and NMI scores of 0.970 even with a $71 \%$ reduction in edges, whereas the Amazon co-purchase network demonstrates consistent performance with a $29 \%$ sparsification. The results emphasize that Metric Backbone sparsification improves computational efficiency and facilitates the use of geodesic-based metrics such as GDM in identifying compact, well-separated communities. This thorough analysis presents the initial systematic review of geodesic-based community evaluation on structurepreserving sparsified networks, delivering valuable insights into the conditions under which parameter-free sparsification preserves rather than obstructs community detection across various network types.
