IEEE Transactions on Computational Social Systems最新文献

{"title":"Unsupervised Textless Prompt With LLMs for Multivariate Time Series Anomaly Detection","authors":"Guoxiang Zhong;Jie Liu;Fagui Liu;Jun Jiang;Bin Wang;Fan Zhang;C. L. Philip Chen","doi":"10.1109/TCSS.2025.3649202","DOIUrl":"https://doi.org/10.1109/TCSS.2025.3649202","url":null,"abstract":"Prompting the pre-trained large language models (LLMs) presents a novel implement to detect multivariate time series anomalies, which demonstrates strong potentiality. However, adding a textual string or embedding to numerical sequence for modeling the prompt has remained elusive due to lacking of exact word descriptions and prior information about initial time series input. In this article, we design an unsupervised textless prompt to fine-tune the LLMs and further propose the LLM-based anomaly detector (LLM-AD). The prompt approach renovates the time-stamp masked operation to variate masked operation based on the observation that the root causes variates about anomalies destroy the normal variates imputation procedure. The unmasked and masked variates are respectively confirmed as prompt input and prompt answer, which gets rid of the text-driven prompt template. Minimizing the divergence between predicted and actual prompt answers fine-tunes the LLMs which only the feedforward and normalization layers are activated. During the next testing stage, an anomaly alert will be triggered if the output answer is made far enough away from actual masked variates. Extensive experiments compared with the nine baseline algorithms reveal that our LLM-AD achieves competitive detection results, boosting the averaged f1-score of four benchmark datasets from 85.28% to 91.62%.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"6390-6402"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097789","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The Impact of Altruism in Symmetric Two-Strategy Population Games","authors":"Colton Hill;Philip N. Brown;Keith Paarporn","doi":"10.1109/TCSS.2026.3713848","DOIUrl":"https://doi.org/10.1109/TCSS.2026.3713848","url":null,"abstract":"It is well known that <italic>selfish</i> behavior, where agents choose strategies to maximize their individual payoff, can lead to degraded social welfare. These inefficiencies can seemingly be mitigated when agents behave <italic>altruistically</i>, and choose strategies intended to increase social welfare. However, in certain settings, altruism can be <italic>perverse</i>, and worsen social welfare compared to if all agents had behaved selfishly. Thus, the goal of this article is to investigate altruistic perversity in the context of symmetric two-strategy population games containing a heterogeneous population of selfish and altruistic agents. We find that if a game has a strictly concave welfare function, then any agent is guaranteed to improve social welfare if they behave altruistically rather than selfishly. Then, we formalize the contrapositive, and show that altruism can be perverse only in population games with a convex social welfare function. Finally, we determine the worst-case harm caused by altruism when the welfare function is convex. Our results provide a complete characterization of how altruism impacts social welfare for the class of symmetric two-strategy games presented in normal form.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"5662-5673"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097986","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing Encrypted Traffic Detection for Social Networks With Inter-Packet Contextual Association","authors":"Xiaowei Zhao;Mingshu He;Xiaojuan Wang;Shize Guo","doi":"10.1109/TCSS.2026.3714853","DOIUrl":"https://doi.org/10.1109/TCSS.2026.3714853","url":null,"abstract":"In social networks, encrypted communication has become increasingly prevalent to protect user privacy. However, this also provides attackers with opportunities to launch covert attacks via encrypted traffic. Traditional traffic detection methods relying on rules and payload analysis fail due to content encryption. As a result, encrypted traffic detection methods based on statistical features and sequence features have gradually emerged as research hotspots. Nevertheless, statistical feature-based methods require access to all packets in a flow, making real-time detection challenging. Therefore, this article exploits sequential features and combines the attention mechanism with sequence prediction to model contextual dependencies among packets. This design allows the model to capture effective flow representations using only a small number of packets, thus helping the classifier achieve accurate classification. In addition, an attention optimization mechanism is introduced to improve the reliability of attention vectors, which contributes to the interpretability of model decisions. Experimental results show that the proposed method achieves competitive classification performance on five datasets using no more than 20 packets, which reduces latency for real-time threat detection. Notably, when using only the first five packets, the average waiting time is only 0.12 s, accounting for merely 0.05% to 3.58% of that required by comparison methods.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"5689-5705"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097024","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Beyond Uniform Modeling: A Dual-View MLP Framework for Trend and Diversity Disentanglement in Multibehavior Recommendation","authors":"Peixuan Sun;Wei Zhou;Min Gao;Xibin Wang;Junhao Wen","doi":"10.1109/TCSS.2025.3630068","DOIUrl":"https://doi.org/10.1109/TCSS.2025.3630068","url":null,"abstract":"In real-world recommendation systems, user interactions contain behavioral information. Multibehavior sequential recommendation focuses on predicting the next item a user will interact with under a specific behavior. Incorporating behavioral information introduces diverse interaction patterns, which we treat as auxiliary signals to supplement trend interest modeling (e.g., primary interests, low-frequency interests, and noise). However, existing multibehavior recommendation methods perform unified modeling without distinguishing the interest categories reflected by different data. Capturing both interest trends and dispersed diversity inherently presents a conflicting objective. Hence, we propose a novel approach that separately models these aspects before integrating them. To realize this vision, we introduce Disentangled Interest Perspective for Multibehavior Sequential Recommendation (DIPRec), which separately captures interest tendencies and diversity before performing a unified integration. The framework begins with a mask generation module that partitions the data, employing multilayer perceptron (MLP)-based modules—specifically, the heterogeneous interest trend module (HITM) and the auxiliary behavior intent perception module (ABIP) to extract user interest tendencies from heterogeneous and auxiliary behavioral sequences. Concurrently, it leverages MLP and max-pooling layers to extract supplementary diversity signals from the separated data. Finally, the model integrates these two aspects to generate recommendations. To assess the effectiveness of our proposed DIPRec model, we conducted thorough experiments on five public datasets. The results consistently demonstrate that DIPRec surpasses current state-of-the-art approaches across a variety of standard evaluation metrics.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"6517-6531"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097145","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Emergence of Cooperation in Spatial Public Goods Games via Grey Wolf Optimization and a Reputation Mechanism","authors":"Yong Shen;Yan Zhang;Hongwei Kang;Xingping Sun;Qingyi Chen;Xinlong Zhang","doi":"10.1109/TCSS.2025.3649080","DOIUrl":"https://doi.org/10.1109/TCSS.2025.3649080","url":null,"abstract":"In public goods games, the core challenge to sustaining cooperation lies in overcoming the temptation for individuals to free-ride, thereby achieving a collective optimum. Classical spatial game models, which typically employ local imitation-based update rules (e.g., the Fermi rule) and discrete binary strategies, have provided a foundational basis for understanding this problem. To explore more complex cooperative dynamics, this study introduces a novel social learning paradigm. Drawing inspiration from the grey wolf optimization (GWO) algorithm, we abstract its mechanism of collaborative hunting by the leadership layer into an active learning process based on local exemplars, which aligns with the real-world behavior of emulating successful peers. This framework transcends the passive imitation of a single strategy; instead, it emphasizes an active learning process where agents identify and synthesize the leadership group within their social network to explore superior cooperative strategies. We further construct a composite fitness metric by integrating long-term reputation with immediate payoffs and adopt a continuous strategy space to enhance behavioral granularity. Numerical simulations demonstrate that this integrated framework—combining active learning, a reputation mechanism, and continuous strategies—significantly promotes the emergence and maintenance of cooperation under stringent conditions. This research not only presents an effective new model for resolving social dilemmas but also offers a more behaviorally realistic theoretical perspective for understanding the evolution of cooperation in complex systems.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"6481-6490"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"KAPT: Knowledge-Augmented Prompt-Tuning Framework for Aspect-Based Sentiment Analysis","authors":"Zhiqiang Zhang;Yiyang Wang;Bing Li;Haiyu Song;Haiyan Wu;Jianyong Wang","doi":"10.1109/TCSS.2026.3708933","DOIUrl":"https://doi.org/10.1109/TCSS.2026.3708933","url":null,"abstract":"Aspect-based sentiment analysis (ABSA) is an important branch of sentiment analysis (SA), primarily used to identify the sentiment polarity of specific aspects or attributes within a sentence. Most studies focus on directly exploring explicit sentiment semantics, neglecting the use of knowledge-enhanced methods to mine implicit textual semantics. To alleviate these problems, we propose a knowledge-augmented prompt-tuning (KAPT) framework. Specifically, we utilize large language models (LLMs) to extract actionable knowledge, thereby overcoming the knowledge constraints inherent in pretrained language models (PLMs). Furthermore, we propose an enhanced prompt architecture and redesigned verbalizers that incorporate the functional features of LLMs to improve the model’s implicit SA capability while optimizing its performance in explicit sentiment processing. Experimental results demonstrate that our model achieves significant performance improvements over state-of-the-art (SOTA) baselines on both Restaurant and Laptop benchmark datasets. Particularly, the accuracy on the laptop dataset is enhanced by 2.04%.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"5745-5759"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097182","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Real-Time Gaining Subscribers by Messaging","authors":"Konstantinos Theocharidis;Hady W. Lauw","doi":"10.1109/TCSS.2026.3708536","DOIUrl":"https://doi.org/10.1109/TCSS.2026.3708536","url":null,"abstract":"The research on how <italic>brands</i> can <italic>gain subscribers using content</i> in their <italic>online pages</i> (maintained in social network and e-commerce platforms) is a recent and interesting service that can be addressed via <italic>messaging</i>. In particular, in each round <inline-formula><tex-math>$boldsymbol{k}$</tex-math></inline-formula> features and <inline-formula><tex-math>$boldsymbol{m}$</tex-math></inline-formula> users are searched so as the <inline-formula><tex-math>$boldsymbol{m}$</tex-math></inline-formula> users to be notified <italic>once</i> by brand advertiser with <italic>messages</i> containing <inline-formula><tex-math>$boldsymbol{k}$</tex-math></inline-formula> features to subscribe to brand page. However, <inline-formula><tex-math>$boldsymbol{(k,m)}$</tex-math></inline-formula>-queries have some noticeable <italic>semantic</i> and <italic>technical</i> limitations. In this article we propose <inline-formula><tex-math>$boldsymbol{k}$</tex-math></inline-formula>-queries to gain subscribers, via using a preference-based estimate of potential subscription gain, to <italic>overcome</i> the limitations of <inline-formula><tex-math>$boldsymbol{(k,m)}$</tex-math></inline-formula>-queries. Specifically, we study the realistic problem of real-time gaining subscribers by messaging (RGSM). RGSM executes in many rounds. In each round, advertiser searches to message <italic>once</i> the <inline-formula><tex-math>$boldsymbol{m}$</tex-math></inline-formula> users who are most interested to <italic>predefined</i> <inline-formula><tex-math>$boldsymbol{k}$</tex-math></inline-formula> features, as in many real-world scenarios, advertiser <italic>already</i> knows (without search) some <inline-formula><tex-math>$boldsymbol{k}$</tex-math></inline-formula> features in the system that are similar to their brand content. The focus on RGSM problem is each <inline-formula><tex-math>$boldsymbol{k}$</tex-math></inline-formula>-query to be answered in <italic>real-time</i> over <italic>many</i> users having <italic>dynamic</i> preferences across rounds. We deploy the algorithms <sc>Scan, Periodicpriority, Periodicrandom, Fixedbuffer</small>, and <sc>Dynamicbuffer</small> to solve RGSM on several <italic>uniform</i> and <italic>nonuniform</i> queries in <italic>real</i> and <italic>synthetic</i> datasets. Results show that overall, <sc>Fixedbuffer</small> and <sc>Dynamicbuffer</small> perform better than other algorithms and constitute the best option for uniform and nonuniform queries, respectively.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"5791-5804"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097184","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Novel Stochastic Fuzzy Epidemic Model With Cognitive Awareness and Reactive Dynamics","authors":"Amir Haji Ali Khamseh’i;Ahmad Khonsari;Masoumeh Moradian","doi":"10.1109/TCSS.2025.3642944","DOIUrl":"https://doi.org/10.1109/TCSS.2025.3642944","url":null,"abstract":"According to the World Health Organization (WHO), infectious diseases are responsible for more than 13 million deaths worldwide each year. Modeling the spread of disease and awareness in a network of individuals is a challenging task that requires accounting for the complex dynamics and interactions of human states and behaviors. A majority of research works in epidemic modeling rely on compartmental models that divide the population into discrete compartments, whereas these models oversimplify the complex dynamics of disease spread. Additionally, traditional models often do not account for the role of dynamic awareness in disease prevention and control, or for the complex interactions between disease and awareness. This article proposes a new general epidemic model that addresses some limitations of the previous models by considering variable values in a continuous state space (CSS). This allows for more realistic and flexible representation of human states and disease transitions. By taking advantage of fuzzy logic, our model addresses the uncertainty and variability of human behavior (as awareness) and reactions to disease propagation (as infection). We also incorporate age-tuned learning and disease dynamics into our model, using Covasim parameters and empirical datasets, capturing age-specific behavioral and epidemiological patterns. The proposed model offers several advantages over traditional compartmental models by providing a more detailed and nuanced representation of the spread of disease and awareness. Also, our age-stratified model adjusts infection and learning rates by age, individual, and neighbor states, capturing nonlinear and stochastic disease and awareness dynamics.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"6326-6341"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097277","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adversarial Optimization of the $delta$-TriBound Objective for Robust Community Detection Attacks Across Multiple Graphs","authors":"Bay-Yuan Hsu;Kuo-Chieh Sun;Ching Chi Yeh","doi":"10.1109/TCSS.2026.3671811","DOIUrl":"https://doi.org/10.1109/TCSS.2026.3671811","url":null,"abstract":"Adversarial attacks on community detection (CD) have attracted increasing attention due to CD’s importance in numerous applications and its privacy implications. While existing studies have demonstrated the effectiveness of adversarial perturbations in degrading CD performance, most prior attacks are restricted to single-graph settings. This limitation contrasts with the growing adoption of multigraph CD frameworks, where community structures are inferred from multiple heterogeneous or cross-platform graphs. Moreover, existing attack methods suffer from two fundamental weaknesses: 1) proxy metrics neglect community structures; and 2) lacking budget allocation ability and the capability to handle multiple graphs simultaneously. To address these challenges, we propose a new research problem, multigraph perturbation for community detection via <inline-formula><tex-math>$delta$</tex-math></inline-formula>-TriBound minimization (<inline-formula><tex-math>$delta$</tex-math></inline-formula>-MPCD). At the core of our formulation is <inline-formula><tex-math>$delta$</tex-math></inline-formula>-TriBound, a proxy metric designed to capture higher-order triangle patterns anchoring community cohesion and to enable effective adversarial perturbations across multiple graphs. We theoretically justify the utility of <inline-formula><tex-math>$delta$</tex-math></inline-formula>-TriBound, prove that <inline-formula><tex-math>$delta$</tex-math></inline-formula>-MPCD is NP-hard, and design a <inline-formula><tex-math>$({1}/{24})$</tex-math></inline-formula>-approximation algorithm, multiple graph <inline-formula><tex-math>$delta$</tex-math></inline-formula>-TriBounds minimization with edge addition (<inline-formula><tex-math>$delta$</tex-math></inline-formula>-MGMA). Extensive experiments on real-world datasets demonstrate that our method consistently degrades community detection performance across diverse CD algorithms in multigraph settings, significantly outperforming state-of-the-art baselines.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"6196-6211"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097303","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"EntroBot: Entropy-Based Modeling of Account Behavior for Social Bot Detection","authors":"Shengyue Jiang;Yuetao Xie;Xiao-Ke Xu;Linyuan Lü","doi":"10.1109/TCSS.2026.3695076","DOIUrl":"https://doi.org/10.1109/TCSS.2026.3695076","url":null,"abstract":"Online social platforms have transformed information dissemination, but their openness has facilitated the rise of social bots that actively influence public discourse at scale. Among existing approaches to social bot detection, behavior-based methods are recognized for robustness and interpretability, though their restricted temporal scope hinders effective modeling of behavioral dynamics and constrains detection accuracy. In this study, we introduce EntroBot, an entropy-based method that detects social bots by modeling account behavior across multiple timescales. Empirical analysis of entropy-quantified behavioral patterns reveals that genuine users maintain behavioral regularity across extended time scales, with stability increasing relative to social bots. Experiments show that EntroBot outperforms existing behavior-driven approaches in detection accuracy and rivals deep learning methods, while preserving interpretability and robustness. Ablation analysis shows that single-scale features are insufficient to characterize diverse bot behaviors, but integrating short-term, long-term, and dynamic signals yields synergistic gains.","PeriodicalId":13044,"journal":{"name":"IEEE Transactions on Computational Social Systems","volume":"13 5","pages":"6048-6061"},"PeriodicalIF":4.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097416","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}