Large-scale online marketplaces and recommender systems are crucial technological foundations for the development of e-commerce. In industrial recommender systems, features play a vital role as they carry essential information for downstream models. Accurate estimation of feature importance is critical, as it helps identify the most useful feature subsets from thousands of candidates for online services. Such a selection enables optimization of online performance while reducing computational burden. To address the feature selection challenges in deep learning, trainable gate-based and sensitivity-based methods have been proposed and proven effective in the industry. However, by analyzing real-world examples, we identified three bias issues that cause feature importance estimation to rely on partial model layers, samples, or gradients, ultimately leading to inaccurate feature importance estimates. We refer to these biases as layer bias, baseline bias, and approximation bias. To mitigate these biases, we propose FairFS, a fair and accurate feature selection algorithm. On one hand, FairFS directly regularizes feature importance estimation across all non-linear transformational layers to avoid layer bias. On the other hand, it employs a smooth baseline feature close to the classifier's decision boundary and an aggregated approximation method to mitigate bias issues. Extensive experiments demonstrate how FairFS mitigates these three biases and achieves state-of-the-art feature selection results.
论文检索
输入标题、作者或关键词，从 5,826 篇学术成果中精准定位
Large Language Models (LLMs) have shown strong potential as conversational agents. Yet, their effectiveness remains limited by deficiencies in robust long-term memory—particularly in complex, long-term Web-based services such as online emotional support. However, existing long-term dialogue benchmarks primarily focus on static and explicit fact retrieval, failing to evaluate agents in these critical scenarios where user information is dispersed, implicit, and continuously evolving. To address this gap, we introduce ES-MemEval, a comprehensive benchmark that systematically evaluates five core memory capabilities—information extraction, temporal reasoning, conflict detection, abstention, and user modeling—in long-term emotional support scenarios, covering question answering, summarization, and dialogue generation tasks. To support the benchmark, we also propose EvoEmo, the first multi-session dataset for personalized long-term emotional support scenarios, capturing fragmented, implicit user disclosures and evolving user states. Extensive experiments on open-source long-context, commercial, and retrieval-augmented (RAG) LLMs reveal that explicit long-term memory is essential to reduce hallucinations and enable effective personalization. At the same time, RAG enhances factual consistency but struggles with temporal dynamics and evolving user states. These findings highlight both the potential and limitations of current paradigms, encouraging the development of more robust memory–retrieval integration in long-term personalized dialogue systems.
As users' preferences evolve over time, personalized online services increasingly rely on sequential recommender systems to predict future interactions by modeling patterns in historical user behavior. However, existing methods for sequential recommendation (SR) face two key challenges: they struggle to simultaneously leverage collaborative, semantic, and rating information, and the use of hard labels during training provides limited supervision. In this paper, we introduce SEAR, an LLM-powered Sequential recommEndation framework via fusion of collAborative, semantic, and Rating information. The proposed deep model comprises an embedding layer and a sequence encoder. The embedding layer transforms user-item interactions into three types of embeddings: collaborative, semantic, and rating. The sequence encoder then integrates these embeddings and identifies sequential patterns to model user representations. To enhance the utilization of item semantics, we integrate a large language model (LLM) to extract LLM embeddings. These embeddings are then employed to initialize the semantic embedding layer, collaborative embedding layer, and item embeddings. To capture more nuanced user behavior patterns, we generate preference-weighted soft labels based on the next k interactions. Extensive experiments validate the effectiveness of SEAR, and ablation studies further highlight the distinct contributions of the collaborative, semantic, and rating information.
With the rapid growth and rising complexity of web content, edge-deployed LLMs have become essential for enhancing users' online experiences. However, sparsely-activated Mixture-of-Experts (MoE) models, which are well-suited for edge scenarios, face significant memory bottleneck challenges. Offload-based methods have been proposed to mitigate the problem, but they face difficulties with expert prediction. To promote the application of MoE models in edge scenarios, we propose Fate, an offloading system designed for MoE models to enable efficient inference in resource-constrained environments. The key insight behind Fate is that gate inputs from adjacent layers can be effectively used for expert prefetching, achieving high prediction accuracy. Furthermore, Fate employs a shallow-favoring expert caching strategy that increases the expert hit rate to 99%. Additionally, Fate integrates tailored quantization strategies for cache optimization and I/O efficiency. Experimental results show that, compared to baselines, Fate achieves up to 1.34×-5.07× prefill speedup and 1.26×–4.41× decoding speedup, while maintaining inference quality.
The rise of Internet of Vehicles (IoV) technologies is transforming traffic management from isolated control to a collective, multi-vehicle process. At the heart of this shift is multi-vehicle dynamic navigation, which requires simultaneously routing large fleets under evolving traffic conditions. Existing path search algorithms and reinforcement learning methods struggle to scale to city-wide networks, often failing to capture the nonlinear, stochastic, and coupled dynamics of urban traffic. To address these challenges, we propose CityNav, a hierarchical, LLM-powered framework for large-scale multi-vehicle navigation. CityNav integrates a global traffic allocation agent, which coordinates strategic traffic flow distribution across regions, with local navigation agents that generate locally adaptive routes aligned with global directives. To enable effective cooperation, we introduce a cooperative reasoning optimization mechanism, in which agents are jointly trained with a dual-reward structure: individual rewards promote per-vehicle efficiency, while shared rewards encourage network-wide coordination and congestion reduction. Extensive experiments on four real-world road networks of varying scales (up to 1.6 million roads and 430,000 intersections) and traffic datasets demonstrate that CityNav consistently outperforms nine classical path search and RL-based baselines in city-scale travel efficiency and congestion mitigation. Our results highlight the potential of LLMs to enable scalable, adaptive, and cooperative city-wide traffic navigation, providing a foundation for intelligent, large-scale vehicle routing in complex urban environments. Our project is available at https://github.com/usail-hkust/CityNav.
Time series anomaly detection (TSAD) has been a long-standing pillar problem in Web-scale systems and online infrastructures, such as service reliability monitoring, system fault diagnosis, and performance optimization. Large language models (LLMs) have demonstrated unprecedented capabilities in time series analysis, the potential of multimodal LLMs (MLLMs), particularly vision-language models, in TSAD remains largely under-explored. One natural way for humans to detect time series anomalies is through visualization and textual description. It motivates our research question: Can multimodal LLMs perform time series anomaly detection? Existing studies often oversimplify the problem by treating point-wise anomalies as special cases of range-wise ones or by aggregating point anomalies to approximate range-wise scenarios. They limit our understanding for realistic scenarios such as multi-granular anomalies and irregular time series. To address the gap, we build a VisualTimeAnomaly benchmark to comprehensively investigate zero-shot capabilities of MLLMs for TSAD, progressively from point-, range-, to variate-wise anomalies, and extends to irregular sampling conditions. Our study reveals several key insights. 1) MLLMs and traditional TSAD methods are complementary: MLLMs excel at coarse-grained anomalies while traditional methods are effective at fine-grained anomalies. 2) MLLMs are resilient to irregular time series. 3) Input time series modality changing from text to image makes information focus shift from quantitative variations to qualitative patterns while significantly reducing hallucinations. Built on the findings, we propose a MLLMs-based multi-agent framework TSAD-Agents to achieve automatic TSAD. Our framework comprises scanning, planning, detection, and checking agents that synergistically collaborate to reason, plan, and self-reflect to enable automatic TSAD. These agents adaptively invoke tools such as traditional methods and MLLMs and dynamically switch between text and image modalities to optimize detection performance.
Traffic prediction serves as a cornerstone for systems and network services such as the Web of Vehicles (WoV), online navigation, and smart city applications. Despite the proliferation of model architectures in recent years, existing approaches often suffer from highly customized structures and weak transferability, making it difficult to cope with increasing task heterogeneity and modeling complexity. To address these challenges, we propose ST-LEGO, a modular assembly framework driven by large language models (LLMs) that supports flexible structural composition and automated code generation. ST-LEGO employs a multi-agent collaborative system comprising a Prompt Agent, Assemble Agent, and Code Agent, which are responsible for understanding task requirements, dynamically assembling structural modules, and automatically generating executable PyTorch code. By introducing a standardized module library and an intermediate structural description language (DSL), the framework enables controllable generation, reusable composition, and cross-task generalization of model architectures. Empirical results on multiple real-world traffic datasets demonstrate that models generated by ST-LEGO achieve superior accuracy, structural diversity, and convergence compared to a wide range of manually designed baselines. These results highlight the unique potential and scalability of LLMs as structural architects for traffic prediction, offering a new paradigm for integrating language models into web-interactive intelligent transportation systems.
Large-scale vision-language models enable powerful cross-modal understanding and generation, driving rapidly growing demand for online inference services. However, cloud-centric serving often suffers from high latency, rising costs, and network dependency, while purely on-device deployment is constrained by limited memory and reduced accuracy on complex tasks. To address this accuracy–latency–cost trilemma, we propose ShiftVL, a task-aware end–cloud serving framework that shifts suitable execution to the end device with a cloud fallback. ShiftVL serves high-frequency requests on an end-side small VLM enhanced with ViTexLoRA, a modality-disentangled parameter-efficient tuning method that preserves cross-modal alignment, while routing low-frequency or complex requests to a cloud-hosted large VLM for higher accuracy. Under tight device budgets, ShiftVL employs a predictive adapter scheduler that combines LRU-style caching with imitation learning to pre-load task-specific adapters. Experiments with InternVL models show that ShiftVL reduces cloud cost by up to 76.3% and latency by up to 42.9% while maintaining high multi-task accuracy, demonstrating its practicality for real-world vision-language model serving.
Influence maximization (IM) aims to select a small set of seed nodes whose activation triggers a maximal cascade. Existing methods typically assume access to the network topology or a reliable surrogate, which is often unavailable in practice due to noisy, privacy-protected, or partially observed links. These settings pose two challenges: (1) the lack of explicit topology removes a key inductive constraint for diffusion modeling, undermining influence estimation, and (2) nonlinear, temporally dependent node interactions yield complex multivariate time series that hinder topology inference. We propose DynaFLUX, an end-to-end generative framework for IM under hidden topology. DynaFLUX learns a compact surrogate of latent dynamics directly from observed time series, and jointly optimizes a seed-selection policy via reinforcement learning. A self-attention pointer network captures long-range dependencies for seed generation, while an influence-prediction module infers a surrogate topology and uses Monte Carlo diffusion to provide policy-gradient rewards. Experiments show that DynaFLUX accurately identifies influential spreaders and consistently outperforms state-of-the-art baselines in unseen topology scenarios.
Online question-and-answer (Q&A) platforms facilitate knowledge exchange through posted questions and answers. Recent advances in large language models (LLMs) have shown their strong capability in generating high-quality answers, leading to a recent surge in LLM-generated answers (LGAs) on Q&A platforms. In this paper, we conduct an in-depth analysis of how LGAs affect Naver Knowledge iN, the most popular Q&A platform in South Korea. To this end, we implement nine state-of-the-art LLM-generated text (LGT) detection methods and evaluate their performance on answers collected from Naver Knowledge iN. We then build an ensemble detector by stacking the three best-performing LGT detection methods, achieving an AUC of 0.9987 with a false positive rate below 1%. Using this LGA detector, we identify 75,558 LGAs among 1.46M answers. We find that LGAs tend to be longer, use more punctuation marks, and exhibit higher lexical diversity. However, LGAs do not show clear differences in user reactions, such as upvotes, downvotes, or selection rates by questioners. We also find that LGAs are primarily intended for knowledge sharing rather than personal experiences sharing. Finally, we observe a shift in the Q&A platform: questions increasingly move from simple fact-seeking to those involving complex contexts and seeking personal opinions or past experiences.
Online narratives spread unevenly across platforms, with content emerging on one site often appearing on others, hours, days or weeks later. Existing cross-platform information diffusion models often treat platforms as isolated systems, disregarding cross-platform activity that might make these patterns more predictable. In this work, we frame cross-platform prediction as a network proximity problem: rather than tracking individual users across platforms or relying on brittle signals like shared URLs or hashtags, we construct platform-invariant discourse networks that link users through shared narrative engagement. We show that cross-platform neighbor proximity provides a strong predictive signal: adoption patterns follow discourse network structure even without direct cross-platform influence. Our highly-scalable approach substantially outperforms diffusion models and other baselines while requiring less than 3% of active users to make predictions. We also validate our framework through retrospective deployment. We sequentially process a datastream of 5.7M social media posts occurred during the 2024 U.S. election, to simulate real-time collection from four platforms (X, TikTok, Truth Social, and Telegram): our framework successfully identified emerging narratives, including crises-related rumors, yielding over 94% AUC with sufficient lead time to support proactive intervention.
Social media engagement prediction is a central challenge in computational social science, particularly for understanding how users interact with misinformation. Existing approaches often treat engagement as a homogeneous time-series signal, overlooking the heterogeneous social mechanisms and platform designs that shape how misinformation spreads. In this work, we ask: ''Can neural architectures discover social exchange principles from behavioral data alone?'' We introduce Dreams (Disentangled Representations and Episodic Adaptive Modeling for Social media misinformation engagements), a social exchange theory-guided framework that models misinformation engagement as a dynamic process of social exchange. Rather than treating engagement as a static outcome, Dreams models it as a sequence-to-sequence adaptation problem, where each action reflects an evolving negotiation between user effort and social reward conditioned by platform context. It integrates adaptive mechanisms to learn how emotional and contextual signals propagate through time and across platforms. On a cross-platform dataset spanning 7 platforms and 2.37M posts collected between 2021 and 2025, Dreams achieves state-of-the-art performance in predicting misinformation engagements, reaching a mean absolute percentage error of 19.25%. This is a 43.6% improvement over the strongest baseline. Beyond predictive gains, the model reveals consistent cross-platform patterns that align with social exchange principles, suggesting that integrating behavioral theory can enhance empirical modeling of online misinformation engagement. The source code is available at: https://github.com/ltian678/DREAMS.
Online hate on social media ranges from overt slurs and threats (hard hate speech ) to soft hate speech: discourse that appears reasonable on the surface but uses framing and value-based arguments to steer audiences toward blaming or excluding a target group. We hypothesize that current moderation systems, largely optimized for surface toxicity cues, are not robust to this reasoning-driven hostility, yet existing benchmarks do not measure this gap systematically. We introduce SoftHateBench, a generative benchmark that produces soft-hate variants while preserving the underlying hostile standpoint. To generate soft hate, we integrate the Argumentum Model of Topics (AMT) and Relevance Theory (RT) in a unified framework: AMT provides the backbone argument structure for rewriting an explicit hateful standpoint into a seemingly neutral discussion while preserving the stance, and RT guides generation to keep the AMT chain logically coherent. The benchmark spans 7 sociocultural domains and 28 target groups, comprising 4,745 soft-hate instances. Evaluations across encoder-based detectors, general-purpose LLMs, and safety models show a consistent drop from hard to soft tiers: systems that detect explicit hostility often fail when the same stance is conveyed through subtle, reasoning-based language. Disclaimer. Contains offensive examples used solely for research.
Opinion formation models are widely used to study social and behavioral processes on online social networks, yet their fairness remains largely unexplored. We study this novel problem in the context of the Friedkin–Johnsen model, a well-established framework for opinion dynamics. In this model, the expressed opinion of an individual evolves by combining peer opinions with a fixed inner opinion weighted by their stubbornness. We define a node's influence as the weight its inner opinion contributes to the public opinion. Given different groups of nodes, we require that influence is distributed fairly across groups. To achieve this, we design minimal interventions that adjust stubbornness, making individuals more receptive to others or more anchored to their own views. We derive closed-form expressions for how changes in the stubbornness of a single node affect influence and leverage them to develop efficient algorithms. Experiments on synthetic and real-world networks provide insights into the role of stubbornness in fairness and demonstrate the effectiveness and efficiency of our methods.
Generative agents are rapidly advancing in sophistication, raising urgent questions about how they might coordinate when deployed in online ecosystems. This is particularly consequential in information operations (IOs), influence campaigns that aim to manipulate public opinion on social media. While traditional IOs have been orchestrated by human operators and relied on manually crafted tactics, agentic AI promises to make campaigns more automated, adaptive, and difficult to detect. This work presents the first systematic study of emergent coordination among generative agents in simulated IO campaigns. Using generative agent-based modeling, we instantiate IO and organic agents in a simulated environment and evaluate coordination across operational regimes, from simple goal alignment to team knowledge and collective decision-making. As operational regimes become more structured, IO networks become denser and more clustered, interactions more reciprocal and positive, narratives more homogeneous, amplification more synchronized, and hashtag adoption faster and more sustained. Remarkably, simply revealing to agents which other agents share their goals can produce coordination levels nearly equivalent to those achieved through explicit deliberation and collective voting. Overall, we show that generative agents, even without human guidance, can reproduce coordination strategies characteristic of real-world IOs, underscoring the societal risks posed by increasingly automated, self-organizing IOs.
Social platforms serve as central hubs for information exchange, where user behaviors and platform interventions jointly shape opinions. However, intervention policies like recommendation and content filtering, can unintentionally amplify echo chambers and polarization, posing significant societal risks. Proactively evaluating the impact of such policies is therefore crucial. Existing approaches primarily rely on reactive online A/B testing, where risks are identified only after deployment, making risk identification delayed and costly. LLM-based social simulations offer a promising pre-deployment alternative, but current methods fall short in realistically modeling platform interventions and incorporating feedback from the platform. Bridging these gaps is essential for building actionable frameworks to assess and optimize platform policies. To this end, we propose PolicySim, an LLM-based social simulation sandbox for the proactive assessment and optimization of intervention policies. PolicySim models the bidirectional dynamics between user behavior and platform interventions through two key components: (1) a user agent module refined via supervised fine-tuning (SFT) and direct preference optimization (DPO) to achieve platform-specific behavioral realism; and (2) an adaptive intervention module that employs a contextual bandit with message passing to capture dynamic network structures. Experiments show that PolicySim can accurately simulate platform ecosystems at both micro and macro levels and support effective intervention policy.
Toxicity detection mitigates the dissemination of toxic content (e.g., hateful comments, posts, and messages within online social actions) to safeguard a healthy online social environment. However, malicious users persistently develop evasive perturbations to disguise toxic content and evade detectors. Traditional detectors or methods are static over time and are inadequate in addressing these evolving evasion tactics. Thus, continual learning emerges as a logical approach to dynamically update detection ability against evolving perturbations. Nevertheless, disparities across perturbations hinder the detector's continual learning on perturbed text. More importantly, perturbation-induced noises distort semantics to degrade comprehension and also impair critical feature learning to render detection sensitive to perturbations. These amplify the challenge of continual learning against evolving perturbations. In this work, we present ContiGuard, the first framework tailored for continual learning of the detector on time-evolving perturbed text (termed continual toxicity detection) to enable the detector to continually update capability and maintain sustained resilience against evolving perturbations. Specifically, to boost the comprehension, we present an LLM powered semantic enriching strategy, where we dynamically incorporate possible meaning and toxicity-related clues excavated by LLM into the perturbed text to improve the comprehension. To mitigate non-critical features and amplify critical ones, we propose a discriminability driven feature learning strategy, where we strengthen discriminative features while suppressing the less-discriminative ones to shape a robust classification boundary for detection. Additionally, we introduce a historical capability replay strategy to preserve previously learned features via feature alignment to alleviate capability forgetting. To the best of our knowledge, this work is the first study on continual toxicity detection against time-evolving evasive perturbed text. Extensive experiments prove the superior performance of ContiGuard over both existing detectors and continual methods. Code and dataset are available at https://github.com/khk-abc/ContiGuard. Warning: This paper contains discussions of harmful content that may be disturbing to some readers.
Detecting social bots is critical to ensuring the security of online discourse and maintaining trust in social networks. Early feature-based and text-based methods often fail against bots that mimic human behavior, and graph-based approaches have emerged to better exploit structural signals. However, most existing Graph Neural Networks (GNNs) still focus on pairwise connections, overlooking higher-order relational patterns, and their multi-relation fusion strategies are typically simplistic, ignoring dependencies between relations and user-specific preferences. To overcome these limitations, we propose MPS-Bot, a model that integrates higher-order structure modeling with user-specific cross-relation dependency learning. MPS-Bot introduces a simplex convolutional layer that leverages simplexes derived from network structures to capture group coordination patterns beyond pairwise connections. In addition, a cross-relation dependency attention mechanism adaptively fuses relation-specific representations according to each user's relational preferences, leading to more discriminative and robust multi-relation representations. Extensive experiments on two widely used Twitter bot detection benchmarks, MGTAB and TwiBot-22, show that MPS-Bot generally outperforms state-of-the-art baselines. These findings highlight the effectiveness of higher-dimensional message passing over simplexes as a powerful approach to unmasking bots in social networks.
Although extensive research has examined social media nudges to influence privacy, comparably less work investigates mechanisms of peer-to-peer influence, or the ways in which peer groups can amplify or dampen privacy considerations. This study takes an experimental approach to examining how people are influenced to share potentially sensitive photos of others without consent---a common practice on social media---by comparing the effects of encouraging, discouraging, or mixed feedback social media comments. The results show that discouraging peer comments reduce the sharing of others' photos when content portrayed targets negatively. However, peer influence backfired among users with a history of online misconduct, with both encouraging and discouraging comments leading to more photo sharing. Moreover, a mix of comments that both discouraged and encouraged sharing led to the highest levels of sharing. Linking to prior work, this research establishes a broader understanding of when privacy nudges are likely to backfire on social media: for users with past online misconduct and when information goes against social norms. The findings advance understanding of how peer-to-peer dynamics shape online information flows and highlight the complex effects of Web-based social feedback on privacy decisions. These insights provide actionable design recommendations for platforms to foster responsible, privacy-conscious sharing norms.
Propagation models are essential for modeling and simulating dynamic processes such as epidemics and information diffusion. However, existing tools struggle to scale to large-scale graphs that emerge across social networks, epidemic networks and so on, due to limited algorithmic efficiency, weak scalability, and high communication overhead. We present FS_GPlib, a unified library that enables efficient, high-fidelity propagation modeling on Web-scale graphs. FS_GPlib introduces a dual-acceleration framework: it combines micro-level synchronous message-passing updates with macro-level batched Monte Carlo simulation, leveraging high-dimensional tensor operations for parallel execution. To further enhance scalability, it supports distributed simulation via a novel target-node-based graph partitioning strategy that minimizes communication overhead while maintaining load balance. Theoretically, we show that under ideal assumptions, the runtime of simulations converges approximately to a constant. Extensive experiments demonstrate up to 35,000× speedup over standard libraries such as NDlib and execution of a full Monte Carlo simulation on a Web-scale (billion-edge) graph in 11 seconds while maintaining high simulation fidelity. FS_GPlib supports 29 propagation models—including epidemic and opinion dynamics and dynamic network models—and offers a lightweight Python API compatible with mainstream data science ecosystems. By addressing the unique challenges of modeling diffusion and cascades on the Web, FS_GPlib provides a scalable, extensible, and theoretically grounded solution for large-scale propagation analysis in epidemiology, social media analysis, and online network dynamics. Code available at: https://github.com/Allen-Ciel/FS_GPlib.