With the rapid growth of Web-based academic publications, more and more papers are being published annually, making it increasingly difficult to find relevant prior work. Citation prediction aims to automatically suggest appropriate references, helping scholars navigate the expanding scientific literature. Here we present CiteRAG, the first comprehensive retrieval-augmented generation (RAG)-integrated benchmark for evaluating large language models on academic citation prediction, featuring a multi-level retrieval strategy, specialized retrievers, and generators. Our benchmark makes four core contributions: (1) We establish two instances of the citation prediction task with different granularity. Task 1 focuses on coarse-grained list-specific citation prediction, while Task 2 targets fine-grained position-specific citation prediction. To enhance these two tasks, we build a dataset containing 7,267 instances for Task 1 and 8,541 instances for Task 2, enabling comprehensive evaluation of both retrieval and generation. (2) We construct a three-level large-scale corpus with 554k papers spanning many major subfields, using an incremental pipeline. (3) We propose a multi-level hybrid RAG approach to citation prediction, fine-tuning embedding models with contrastive learning to capture complex citation relationships, paired with specialized generation models. (4) We conduct extensive experiments across state-of-the-art language models, including closed-source APIs, open-source models, and our fine-tuned generators, demonstrating the effectiveness of our framework. Our open-source toolkit enables reproducible evaluation and focuses on academic literature, providing the first comprehensive evaluation framework for citation prediction and serving as a methodological template for other scientific domains. Our source code and data are released at https://github.com/LQgdwind/CiteRAG.
论文检索
输入标题、作者或关键词，从 12,319 篇学术成果中精准定位
Misinformation and fake news have become a pressing societal challenge, driving the need for reliable automated detection methods. Prior research has highlighted sentiment as an important signal in fake news detection, either by analyzing which sentiments are associated with fake news or by using sentiment and emotion features for classification. However, this poses a vulnerability since adversaries can manipulate sentiment to evade detectors especially with the advent of large language models (LLMs). A few studies have explored adversarial samples generated by LLMs, but they mainly focus on stylistic features such as writing style of news publishers. Thus, the crucial vulnerability of sentiment manipulation remains largely unexplored. In this paper, we investigate the robustness of state-of-the-art fake news detectors under sentiment manipulation. We introduce AdSent, a sentiment-robust detection framework designed to ensure consistent veracity predictions across both original and sentiment-altered news articles. Specifically, we (1) propose controlled sentiment-based adversarial attacks using LLMs, (2) analyze the impact of sentiment shifts on detection performance. We show that sentiment change heavily impacts the performance of detection models, indicating biases towards neutral articles being real, while non-neutral articles are often classified as fake. (3) We introduce a novel sentiment-agnostic training strategy that enhances robustness against such perturbations. Extensive experiments on three benchmark datasets demonstrate that AdSent significantly outperforms competitive baselines in both accuracy and robustness, while also generalizing effectively to unseen datasets and adversarial scenarios.
IConMoE: Modeling Intents of Misinformation using Concept Activation Vector-based Mixture of Experts
Understanding the intent behind misinformation helps reveal distinct patterns and underlying motives, enabling more accurate and efficient detection. However, existing approaches often struggle to effectively model the interplay of multiple coexisting intents within claims, and typically require additional LLM inferences to infer intent for each sample. In this paper, we introduce an intent-informed concept activation vector-based mixture-of-experts framework, iConMoE, which models latent intents in an unsupervised manner and predicts the veracity of claims in an end-to-end fashion, without the need for additional LLM inferences. Evaluated on three benchmark datasets, our method consistently outperforms both baseline and state-of-the-art models. We also conduct comprehensive ablation studies and error analyses to highlight the contributions of individual components and identify key challenges. In addition, we provide a qualitative analysis of the learned intents across samples to offer deeper insights into the model's reasoning.
Detecting harmful memes is crucial for safeguarding the integrity and harmony of online environments, yet existing detection methods are often resource-intensive, inflexible, and lacking explainability, limiting their applicability in assisting real-world web content moderation. We propose U-CoT+, a resource-efficient framework that prioritizes accessibility, flexibility and transparency in harmful meme detection by fully harnessing the capabilities of lightweight unimodal large language models (LLMs). Instead of directly prompting or fine-tuning large multimodal models (LMMs) as black-box classifiers, we avoid immediate reasoning over complex visual inputs but decouple meme content recognition from meme harmfulness analysis through a high-fidelity meme-to-text pipeline, which collaborates lightweight LMMs and LLMs to convert multimodal memes into natural language descriptions that preserve critical visual information, thus enabling text-only LLMs to ''see'' memes by ''reading''. Grounded in textual inputs, we further guide unimodal LLMs' reasoning under zero-shot Chain-of-Thoughts (CoT) prompting with targeted, interpretable, context-aware, and easily obtained human-crafted guidelines, thus providing accountable step-by-step rationales, while enabling flexible and efficient adaptation to diverse sociocultural criteria of harmfulness. Extensive experiments on seven benchmark datasets show that U-CoT+ achieves performance comparable to resource-intensive baselines, highlighting its effectiveness and potential as a scalable, explainable, and low-resource solution to support harmful meme detection.
Current contrastive graph clustering approaches suffer from insufficient integration of structural and semantic information, coupled with the absence of reliable sample selection strategies. To address these dual limitations, we introduce Structure-Semantic Synergized Deep Contrastive Graph Clustering (S³-DCGC), a novel framework that jointly models topological structure and semantic features through two synergistic mechanisms. First, a structure-aware curriculum negative sampling strategy progressively identifies hard negative samples using dynamic-range masking, enhancing discriminative structural learning. Second, a semantic confidence-guided contrastive mechanism quantifies node reliability via composite confidence scores—integrating cluster affinity and cross-view consistency—to select high-confidence positive/negative pairs. Dynamically coordinated by a soft-alignment strategy that shifts optimization focus from structural to semantic dominance during training, these components achieve balanced synergy. Comprehensive experiments conducted on five benchmark datasets demonstrate S³-DCGC's superiority, achieving significant performance gains. Ablation studies and visual analyses further corroborate the critical importance of structure-semantic synergy in achieving robust clustering performance.
Multi-Agent Systems (MAS) offer a powerful paradigm for solving complex problems, yet their performance is critically dependent on the design of their underlying collaboration topology. As MAS become increasingly deployed in web services (e.g., search engines), designing adaptive topologies for diverse cross-domain user queries becomes essential. Current graph learning-based design methodologies often adhere to a ''one-for-one'' paradigm, where a specialized model is trained for each specific task domain. This approach suffers from poor generalization to unseen domains and fails to leverage shared structural knowledge across different tasks. To address this, we propose OFA-MAS, a one-for-all framework that generates adaptive collaboration graphs for any task described in natural language through a single universal model. Our approach integrates a Task-Aware Graph State Encoder (TAGSE) that filters task-relevant node information via sparse gating, and a Mixture-of-Experts (MoE) architecture that dynamically selects specialized sub-networks to drive node and edge prediction. We employ a three-stage training strategy: unconditional pre-training on canonical topologies for structural priors, large-scale conditional pre-training on LLM-generated datasets for task-topology mappings, and supervised fine-tuning on empirically validated graphs. Experiments across six diverse benchmarks show that OFA-MAS significantly outperforms specialized one-for-one models, generating highly adaptive MAS topologies. Code: https://github.com/Shiy-Li/OFA-MAS.
Graph Neural Networks (GNNs) are vulnerable to graph backdoor attacks, which poses severe risks to their deployment in safety-critical applications. Existing defenses predominantly focus on specific backdoor triggers, making them brittle and unable to generalize across different backdoor triggers with varying properties. Motivated by this limitation, this work proposes an attack-agnostic graph backdoor defense mechanism A2GBD, which does not require prior knowledge of the specific attack strategies (e.g., edge perturbation, node attribute manipulation) to achieve effective defense. A2GBD consists of suspicious node selection and defense strategy generation. The selection module selects high-suspicion nodes to enhance defense awareness, while the defense agent adaptively determines and executes defense strategies. Extensive experiments on multiple benchmark datasets demonstrate that A2GBD consistently lowers attack success rates while maintaining high clean accuracy, showing strong robustness and generalizability against diverse graph backdoor attack strategies.
Graph anomaly detection identifies abnormal patterns in networks but faces label scarcity and extreme class imbalance. While graph contrastive learning offers unsupervised solutions, existing methods suffer from two limitations: random augmentations break semantic consistency in positive pairs, while naive negative sampling produces trivial contrasts. We propose AC2L-GAD, an Active Counterfactual Contrastive Learning framework addressing both limitations through principled counterfactual reasoning. By combining information-theoretic active selection with counterfactual generation, our approach identifies structurally complex nodes and generates anomaly-preserving positive augmentations alongside hard negative contrasts, while restricting expensive counterfactual generation to a strategically selected subset. This design reduces computational overhead by approximately 65% compared to full-graph counterfactual generation while maintaining detection quality. Experiments on nine benchmark datasets, including real-world financial transaction graphs from GADBench, show that AC2L-GAD achieves competitive or superior performance compared to state-of-the-art baselines, with notable gains in datasets where anomalies exhibit complex attribute-structure interactions.
Graph Neural Networks (GNNs) have become the dominant paradigm for machine learning on relational data, yet they remain susceptible to overfitting and noise in graph structures. While data augmentation has proven effective for regularization across domains, existing graph methods operate at coarse levels, such as perturbing entire structures or mixing node features. These approaches are context-agnostic and do not target the core computational process of GNNs: message passing. We introduce MessageShift, a novel fine-grained data augmentation paradigm that operates directly on the messages, the atomic units of information, as they flow through the GNN. The core idea is to apply a contextual perturbation to each message by shifting it towards or away from the center of its local neighborhood. This provides a rich regularization effect, capable of both smoothing noisy messages and sharpening distinctive ones. Extensive experiments on a wide range of benchmark datasets demonstrate that MessageShift consistently outperforms strong baselines across multiple GNN backbones.
Multi-Graph Learning (MGL) is a fundamental machine learning paradigm that represents objects as bags-of-graphs, each encoding a distinct structural property, and has broad applications in bioinformatics, chemistry, computing power networks, and software defect detection. However, the inherent scarcity of labeled data poses a significant bottleneck for supervised MGL approaches. While self-supervised contrastive learning offers a compelling solution, its direct application to MGL faces three key challenges: (1) existing graph neural networks, primarily for single-graph modeling, struggle to yield discriminative bag-level representations from bags-of-graphs; (2) conventional contrastive objectives are limited to single-level settings, failing to capture cross-hierarchical dependencies; and (3) standard data augmentation often disrupts intrinsic graph and bag structures, undermining semantic consistency. To address these issues, we propose the Hierarchical Graph-Bag-Network (HGBN), a self-supervised MGL framework that constructs hierarchical representations in the form of a graph-bag-network. HGBN employs an asymmetric hierarchical graph neural network to learn discriminative graph-level and bag-level representations, introduces cross-hierarchical contrastive objectives to align graph-level and bag-level semantics, and leverages the asymmetric network outputs to form positive and negative pairs, preserving intrinsic structural and semantic consistency. Experiments on eight benchmark multi-graph datasets demonstrate that HGBN consistently outperforms both supervised and self-supervised state-of-the-art baselines, achieving average improvements of 4.82% in accuracy and F1 score.
Social event detection (SED) enhances public awareness by clustering large-scale social messages and has been widely applied across diverse domains. While many existing frameworks adopt Graph Convolution Networks (GCN) as backbones and extend them with auxiliary modules to model complex message relationships, they rarely explore the deeper potential of GCN itself. Specifically, the core operation of GCN—smoothed feature aggregation—implicitly assumes that social information diffusion follows a single-pass and independent heat propagation process. However, real-world diffusion is inherently multi-wave and oscillatory, where different messages interact through reinforcement and interference, resembling the principle of wave–particle duality. To address this gap, we propose Physics-Inspired Graph Convolution Networks (PIGCN ), a novel model that unifies wave-based propagation and particle-like interactions. Specifically, PIGCN integrates a Helmholtz Graph Filter module to capture spectral wave propagation with oscillatory dynamics, and a Physical Interaction Force mechanism to adaptively adjust edge weights by attracting messages within the same event and repelling those across events. Building on PIGCN, we further incorporate a contextualized text encoder and timestamp encoding to form a comprehensive SED framework. Extensive experiments demonstrate that PIGCN effectively outperforms conventional GCN baselines and achieves superior performance across multiple SED datasets, establishing a new benchmark for balancing architectural simplicity with effectiveness. Our code can be found on GitHub . https://github.com/yuyongsheng1990/PIGCN
Semi-supervised graph anomaly detection confronts the fundamental challenge of identifying anomalous nodes that exhibit deviations from normal graph patterns in node attributes or structure connectivity, using a small set of labeled normal nodes. Existing methods fall short in modeling the spectrum of anomaly severity, as they generate anomalies in a single, undifferentiated step. This oversight restricts the detection of complex anomalies. In this paper, we propose a progressive anomaly generation network (PAGE) to overcome this limitation. PAGE enhances complex anomaly detection through progressive attribute-structure perturbation: It first injects attribute noise into selected normal nodes to generate pseudo-anomalous nodes. Then it applies progressive structure perturbation to create hybrid anomalies that simulate the evolution from mild to complex perturbations. Importantly, the proposed anomaly progressive constraint loss enforces that hybrid anomalies exhibit a higher degree of abnormality than primary anomalies, enhancing the model's anomaly quantification capability. PAGE further integrates deviation constraint loss and classification loss to optimize node representations and improve prediction. Extensive experiments on benchmark datasets demonstrate that PAGE significantly outperforms state-of-the-art methods, demonstrating its effectiveness for graph anomaly detection.
Graph-level anomaly detection (GLAD) is a critical task to identify graphs with abnormal properties in various domains, ranging from fraudulent social networks to malicious botnets on online platforms. The dominant paradigm for existing GLAD detectors has been partially unsupervised, relying on training data composed exclusively of normal samples. However, this partially unsupervised paradigm inevitably requires a costly expert filtering process to ensure the training data is free of anomalies. This creates a significant gap between current approaches and the real-world necessity of a fully unsupervised paradigm, which involves training a model directly on real-world data ''as-is'', with its inherent mix of normal and anomalous samples. To bridge this gap, we incorporate uncertainty learning into GLAD to promote fully unsupervised learning. We propose two frameworks: Score Uncertainty Learning (SUL) and Graph-data Uncertainty Learning (GUL). Specifically, SUL enhances existing GLAD detectors by modeling uncertainty through Gaussian distributions over the detectors' predictions, adaptively attenuating the influence of potential anomalies. GUL is an end-to-end framework that iteratively optimizes anomaly detection and uncertainty modeling via an Expectation-Maximization algorithm. In addition, we develop a dedicated loss that utilizes potential anomalies to enhance the effectiveness and robustness of GUL. Empirical results on sixteen benchmark datasets, covering real-world graphs from social networks and online platforms, demonstrate the superiority of our methods and highlight the promise of incorporating uncertainty into fully unsupervised GLAD.
Subgraph Anomaly Detection (SAD) is crucial for identifying groups that deviate from the regular pattern within graphs, which benefits different domains such as financial fraud and network security. However, current studies rely on traditional node detection methods and fixed sampling strategies of subgraph structures, which makes it difficult to learn the pattern collaboration behavior of subgraphs. To address this limitation, this paper proposes a novel unsupervised framework named PC-SAD. The PC-SAD framework first employs an improved Graph AutoEncoder to identify core anomaly nodes by capturing multi-scale neighborhood information. Starting from these core anomaly nodes, we sample candidate subgraphs with path, tree, and cyclic structures, and enhance them according to the characteristics of the subgraph structures. Subsequently, candidate subgraphs are fed into the proposed Pattern Collaboration-based Graph Contrastive Learning method to generate collaborative pattern embeddings, thereby distinguishing anomaly subgraphs. The experimental results show that PC-SAD outperforms the state-of-the-art baseline methods on four benchmark datasets, which proves that PC-SAD is an effective solution to detect anomaly subgraphs.
Dynamic graph representation learning has become essential for analyzing evolving networks in domains such as social network analysis, recommendation systems, and traffic analysis. However, existing continuous-time methods face three key challenges: (1) some methods depend solely on node-specific memory without effectively incorporating information from neighboring nodes, resulting in embedding staleness; (2) most fail to explicitly capture correlations between node neighborhoods, limiting contextual awareness; and (3) many fail to fully capture the structural dynamics of evolving graphs, especially in absence of rich link attributes. To address these limitations, we introduce TAWRMAC—a novel framework that integrates Temporal Anonymous Walks with Restart, Memory Augmentation, and Neighbor Co-occurrence embedding. TAWRMAC enhances embedding stability through a memory-augmented GNN with fixed-time encoding and improves contextual representation by explicitly capturing neighbor correlations. Additionally, its Temporal Anonymous Walks with Restart mechanism distinguishes between nodes exhibiting repetitive interactions and those forming new connections beyond their immediate neighborhood. This approach captures structural dynamics better and supports strong inductive learning. Extensive experiments on multiple benchmark datasets demonstrate that TAWRMAC consistently outperforms state-of-the-art methods in dynamic link prediction and node classification under both transductive and inductive settings across three different negative sampling strategies. By providing stable, generalizable, and context-aware embeddings, TAWRMAC advances the state of the art in continuous-time dynamic graph learning.
Graph Foundation Models (GFMs) have emerged as a frontier in graph learning, which are expected to deliver transferable representations across diverse tasks. However, GFMs remain constrained by in-memory bottlenecks: they attempt to encode knowledge into model parameters, which limits semantic capacity, introduces heavy lossy compression with conflicts, and entangles graph representation with the knowledge in ways that hinder efficient adaptation, undermining scalability and interpretability. In this work,we propose RAG-GFM, a Retrieval-Augmented Generation aided Graph Foundation Model that offloads knowledge from parameters and complements parameterized learning. To externalize graph knowledge, we build a dual-modal unified retrieval module, where a semantic store from prefix-structured text and a structural store from centrality-based motif. To preserve heterogeneous information, we design a dual-view alignment objective that contrasts both modalities to capture both content and relational patterns. To enable efficient downstream adaptation, we perform in-context augmentation to enrich supporting instances with retrieved texts and motifs as contextual evidence. Extensive experiments on five benchmark graph datasets demonstrate that RAG-GFM consistently outperforms 13 state-of-the-art baselines in both cross-domain node and graph classification, achieving superior effectiveness and efficiency.
Accurate traffic flow forecasting is crucial for intelligent transportation systems and relies on effectively modeling complex spatio-temporal dependencies. Although recent graph-based deep learning methods have achieved promising results, most focus on pairwise neighbor relationships, limiting their ability to capture higher-order spatio-temporal interactions in the traffic network. To overcome this limitation, we propose a novel Multi-source information driven Spatio-Temporal HyperGraph learning for traffic forecasting (MSTHG), which is designed to capture richer relational and semantic information. MSTHG introduces a multi-source hypergraph fusion strategy that jointly models dynamic high-order spatial and temporal correlations. Specifically, we build a spatial hypergraph based on geographical proximity to represent high-order spatial dependencies, and a temporal-trend hypergraph leveraging mutual information to capture nonlinear similarities among traffic series. To enhance the semantic richness of node representations, we integrate key daily and weekly information along with periodic features derived from Fast Fourier Transform (FFT). Following the obtained hypergraph, node representations are learned through a hypergraph convolutional network and subsequently processed by a GRU-MLP fusion module, which is designed to capture both local and global temporal dependencies. Extensive experiments on real-world benchmark datasets demonstrate that MSTHG outperforms state-of-the-art baselines. The source code is https://github.com/April-leng/MSTHG.git.
The ''pre-train, prompt'' paradigm, designed to bridge the gap between pre-training tasks and downstream objectives, has been extended from the NLP domain to the graph domain and has achieved remarkable progress. Current mainstream graph prompt-tuning methods modify input or output features using learnable prompt vectors. However, existing approaches are confined to single-granularity (e.g., node-level or subgraph-level) during prompt generation, overlooking the inherently multi-scale structural information in graph data, which limits the diversity of prompt semantics. To address this issue, we pioneer the integration of multi-scale information into graph prompt and propose a Multi-Scale Graph Chain-of-Thought (MSGCOT) prompting framework. Specifically, we design a lightweight, low-rank coarsening network to efficiently capture multi-scale structural features as hierarchical basis vectors for prompt generation. Subsequently, mimicking human cognition from coarse-to-fine granularity, we dynamically integrate multi-scale information at each reasoning step, forming a progressive coarse-to-fine prompt chain. Extensive experiments on eight benchmark datasets demonstrate that MSGCOT outperforms the state-of-the-art single-granularity graph prompt-tuning method, particularly in few-shot scenarios, showcasing superior performance. The code is available at: https://github.com/zhengziyu77/MSGCOT.
In this paper, we propose Graph Retention Networks (GRNs) as a unified architecture for deep learning on dynamic graphs. The GRN extends the concept of retention into dynamic graph data as graph retention, equipping the model with three key computational paradigms: parallelizable training, low-cost O(1) inference, and long-term chunkwise training. This architecture achieves an optimal balance between efficiency, effectiveness, and scalability. Extensive experiments on benchmark datasets demonstrate its strong performance in both edge-level prediction and node-level classification tasks with significantly reduced training latency, lower GPU memory overhead, and improved inference throughput by up to 86.7x compared to SOTA baselines. The proposed GRN architecture achieves competitive performance across diverse dynamic graph benchmarks, demonstrating its adaptability to a wide range of tasks.
High dialogue engagement is a crucial indicator of an effective conversation. A reliable measure of engagement could help benchmark large language models, enhance the effectiveness of human-computer interactions, or improve personal communication skills. However, quantifying engagement is challenging, since it is subjective and lacks a ''gold standard''. This paper proposes PMIScore, an efficient unsupervised approach to quantify dialogue engagement. It uses pointwise mutual information (PMI), which is the probability of generating a response conditioning on the conversation history. Thus, PMIScore offers a clear interpretation of engagement. As directly computing PMI is intractable due to the complexity of dialogues, PMIScore learned it through a dual form of divergence. The algorithm includes generating positive and negative dialogue pairs, extracting embeddings by large language models (LLMs), and training a small neural network using a mutual information loss function. We validated PMIScore on both synthetic and real-world datasets. Our results demonstrate the effectiveness of PMIScore in PMI estimation and the reasonableness of the PMI metric itself.