Evaluating retrieval-augmented generation (RAG) systems in legal domains is challenging due to the nuanced nature of legal reasoning and scarcity of domain-specific benchmarks. We investigate semantic similarity and legal correctness in Brazilian legal question answering, developing a synthetic dataset of 3,012 evaluation instances from the Brazilian Civil Procedure Code spanning seven query types, validated through human expert assessment. Our evaluation framework combines BERTScore with domain-adapted LLM-as-Judge (GPT-4o-mini), validated against expert legal assessment. Analysis reveals 50.8% disagreement between semantic similarity and legal correctness, with correlation varying substantially by query type (ρ = 0.464 to ρ = 0.732). Explicit article citation emerges as the strongest quality predictor (Cohen’s d = 1.099), with cited responses achieving 145% higher legal correctness scores. Concrete examples demonstrate bidirectional divergence: responses may achieve high semantic similarity yet cite incorrect articles, or provide legally perfect answers with minimal lexical overlap. Findings demonstrate that semantic metrics alone are insufficient for legal RAG evaluation, with disagreement patterns structured by query type.
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
Large Language Models (LLMs) achieve strong performance in many applications but remain limited in handling graph-structured data due to their reliance on textual context. Recent approaches integrate Graph Neural Networks (GNNs) to enhance structural modeling, yet they largely overlook fairness, leaving models vulnerable to bias amplification across graph and text modalities. To address this issue, we propose FairGEnt, a disentangled graph-enhanced large language model for fair graph learning. FairGEnt separates sensitive-related and sensitive-invariant factors in both graph and textual representations to mitigate bias while preserving task-relevant information, and further aligns the two modalities through a fairness-aware integration module. In addition, FairGEnt incorporates fair graph-enhanced instruction tuning to improve LLM understanding of complex graph structures. Experiments on multiple benchmark datasets demonstrate that FairGEnt consistently outperforms existing methods in both fairness and predictive performance.
Attributing model behavior to training data is an evolving research field. A common benchmark is data removal, which involves eliminating data instances with either low or high values, then assessing a model's performance trained on the modified dataset. Many existing studies leverage Shapley-based data values for this task. In this paper, we demonstrate that these data values are not optimally suited for pruning low-value data when only a limited amount of data remains. To address this limitation, we introduce the Constraint-Data-Value-Maximization (CDVM) approach, which effectively utilizes data attributions for pruning in low-data scenarios. By casting pruning as a constrained optimization that both maximizes total influence and penalizes excessive per-test contributions, CDVM delivers robust performance when only a small fraction of the data is retained. On the OpenDataVal benchmark, CDVM shows strong performance and competitive runtime.
Emotion and intent joint understanding in multimodal conversations (MC-EIU) aims to infer emotion and intent information by leveraging the semantic dependencies among multimodal data. However, prior works largely ignore redundancy interference and suffer from insufficient interaction and inter-task noise propagation due to their reliance on shallow mechanisms. To overcome these limitations, we propose a novel framework named Disentangled Hierarchical Interaction and Adaptive Adjustment (DELTA) for MC-EIU. We first design a Disentangled Feature Denoising module based on orthogonal decomposition to effectively filter out noise and redundancy from heterogeneous data. Second, we propose a Dual-Level Adaptive Adjustment mechanism to dynamically optimize learning dynamics from both modality and instance perspectives. Furthermore, we introduce a Hierarchical Task-Bottleneck Interaction(HTBI) module, which employs a set of progressively compressed bottleneck tokens as a communication hub. This design can effectively simulating the multi-round iterative interaction between emotion and intent tasks. Experiments on the benchmark MC-EIU bilingual dataset demonstrate that our framework significantly outperforms state-of-the-art baselines in both emotion and intent tasks.
Deep multi-view clustering draws plentiful attention in various domains, owing to remarkable performance in learning patterns from complementary information of multi-view data. However, previous methods encounter two challenges. They utilize a single pre-defined clustering strategy to perceive diverse structures from data in multiple views subject to heterogeneous distributions, failing to fully capture intricate complementary structures. They leverage either the feature fusion or the result fusion in clustering, which cannot fully integrate complementary information. Therefore, a decoupled multi-agent actor-critic (PRIME) is proposed via defining multi-view clustering as a partially observable Markov game, which establishes a dynamic trial-and-error assignment process of samples in a decentralized perception with centralized learning framework to adaptively learn the optimal clustering strategy. In PRIME, an actor leverages the policy gradient paradigm to independently implement a Markov decision process of data partition in a view, which fully explores data structures to tailor a clustering sub-policy for a view. Meanwhile, a critic utilizes the value function paradigm to centrally guide the Markov game among actors in different views, which constructs both feature and result fusion to progressively enhance complementary knowledge integration for robust clustering results. Extensive experiments on 6 benchmark datasets verify the superiority of PRIME against 10 methods.
Accurate long-term forecasting in complex systems is frequently compromised by dataset-level distribution shifts, where diverse underlying behavioral modes and evolving system states drive the dynamic multivariate time-series. While existing methods predominantly focus on local temporal shifts, they fail to explicitly model the global structural challenge where datasets are composites of distinct operational regimes. In this paper, we propose NEST, a specialized framework designed to model and recompose these evolving structures through a two-phase dense MoE architecture. NEST first facilitates structural specialization by partitioning the dataset into distinct operational regimes through unsupervised clustering in a principled moment-entropy space. We introduce a regime-oriented router mechanism that generates initial expert weights based on temporal content, subsequently refined through geometric modulation to regime centroids. Crucially, rather than acting as monolithic predictors, individual experts function as specialized kernels that capture regime-specific dynamics by evolving unique variate-attention patterns. Extensive evaluations on diverse benchmarks, including heterogeneous network traffic and physical phenomena, demonstrate that NEST consistently achieves state-of-the-art performance. Our code and datasets are available at \url{https://github.com/Aaralshin/NEST}.
Modeling emotional dynamics in multi-turn dialogues remains challenging due to emotion shifts, where emotional states evolve gradually rather than change abruptly. Existing methods often rely on one-hot supervision, which fails to reflect the progressive nature of human emotions, particularly for neutral-valence emotions with subtle trajectories. To address this limitation, we propose a pseudo soft-label guided Progressive Hybrid-Distributional (PHD) learning framework. PHD reconstructs discrete labels into hybrid distributional representations that encode inter-emotion relations and mixed emotional tendencies. Based on these representations, a progressive training strategy is introduced to guide the model from learning blended emotional states toward the standard one-to-one prediction objective. Furthermore, we design a Graded Contrastive Learning mechanism that replaces rigid binary allocation with graded supervision to alleviate label conflicts. Experiments on benchmark datasets demonstrate PHD consistently improves diverse baseline models, yielding average w-F1 gains of 1.29%. Our framework highlights the critical importance of modeling emotional dynamics as a gradual, distributional process.
Modeling inter-channel dependencies is important for multivariate time series forecasting (MTSF). However, in many cases, inter-channel dependencies are time-varying and subject to noise interference, making it difficult for models to find a balance between structural stability and temporal adaptivity. Existing methods either use a single global static structure, resulting in insufficient sensitivity to temporal changes, or use local statistical correlations to construct dependencies, but local correlations are prone to introducing noise, which may further amplify the impact of noise during propagation. To address these issues, we propose a Constraint-Guided Dynamic Graph Network (ConDyGNet), whose core idea is "global basis, dynamic weights". Specifically, ConDyGNet learns a low-rank global basis as a shared structural constraint and generates patch-wise basis mixing weights to construct dynamic propagation graphs. This maintains topological consistency while allowing local adaptation and reducing the influence of local noise. We conducted extensive experiments on eight public benchmark datasets and multiple forecasting horizons, demonstrating that ConDyGNet can learn more robust time-varying inter-channel dependencies and achieve state-of-the-art forecasting accuracy. The code is available at https://github.com/constli67/ConDyGNet.
Masked autoencoder have demonstrated excellent performance in point cloud representation learning and received widespread attention. However, due to the unstructured nature of point clouds, existing methods struggle to effectively model both local geometric information and global topological features, while generally neglecting positional information, resulting in insufficient model representational capacity. To address these challenges, we propose a Position-Assigned Graph Masked Autoencoder (PA-GMAE) framework for point cloud representation learning. Specifically, we first explicitly transform the unstructured point clouds into a graph structure and assigns it positional information, leveraging the structured properties of graph to mine the intrinsic correlations among points. During the encoding stage, global contextual are dynamically injected into node features, endowing the model with capabilities of global awareness. In the decoding stage, collaborative work across multiple decoding tasks fully captures the global topology and positional dependencies of the point cloud, enhancing the quality of learned representations. Extensive experiments are conducted on three authoritative benchmark datasets, namely ScanObjectNN, ModelNet40, and ShapeNetPart, across typical downstream tasks including point cloud classification and part segmentation, the results demonstrate the effectiveness and superiority of the proposed framework.
Split Federated Learning (SFL) has emerged as a pivotal paradigm for privacy-preserving distributed training on resource-constrained edge devices by partitioning neural networks between clients and a server. A critical design choice in SFL is the split layer, which determines the computation distribution and the semantic level of smashed data, directly impacting communication overhead, training latency, and model accuracy. Existing SFL methods largely ignore the time-varying nature of edge resources, relying on static resource profiles that lead to suboptimal efficiency and compromise model convergence. In this paper, we propose Dynamic Adaptive Split Federated Learning (DASFL), a unified framework that jointly addresses dynamic resource heterogeneity and non-IID data distributions. We employ a resource-aware dynamic split layer selection strategy that enables each client to minimize per-round latency by adapting to instantaneous local conditions. We also design an adaptive masked aggregation that robustly synchronizes client updates under split-induced structural heterogeneity. Extensive experiments on multiple benchmark models and datasets, conducted under heterogeneous, time-varying resource conditions and non-IID data distributions, demonstrate that DASFL achieves a superior accuracy-efficiency trade-off and faster convergence compared to state-of-the-art SFL baselines.
Knowledge concept tagging aims to assign specific concept or topic labels to educational content, which is essential for both educators and learners in traditional and online teaching practices. Recent work has explored large language models (LLMs) for this task, achieving promising performance. However, LLMs still struggle to select the correct concept from a large-scale candidate set due to the high dimensionality of the decision space. In this paper, we propose a novel three-stage Select-Reason-Judge (SRJudge) framework, which empowers LLMs with selective reasoning capability for fine-grained knowledge concept tagging. Specifically, the Selector in Stage 1 first narrows the candidate concepts to a top-K shortlist by fine-tuning a small language model (SLM), e.g., BERT, since the top-K predictions hit the correct concept in most cases, thereby reducing the decision space of correct candidates. Next, the Stage 2 Reasoner employs a lightweight LLM for refined reasoning over the shortlisted candidates. It further integrates an improved reinforcement learning strategy with a dynamic task-specific reward function and a pruning mechanism to better align with human reasoning preferences. Finally, a larger LLM acts as a judger that evaluates the overall rationality of the reasoning process and its explanations to determine the final output. In addition, we construct two high-quality datasets for further validation, i.e., the biology dataset S_Bio and the physics dataset S_Phy. Experimental results demonstrate that our method consistently outperforms state-of-the-art baselines across benchmark datasets, verifying its effectiveness and superiority. Resources are available at: https://github.com/Nicozwy/SRJudge.
Graph anomaly detection (GAD) aims to identify nodes that exhibit significant deviations from expected structural or attribute patterns, and has garnered increasing attention in recent years. Recent approaches for GAD have predominantly focused on local inconsistency mining, which refers to the difficulty of establishing high similarity relationships between anomalous nodes and their neighbors. While local inconsistency mining requires the incorporation of topological information, the use of Graph Neural Networks (GNNs) to introduce such information tends to homogenize connected nodes, thereby causing the loss of local anomalous signals. To address this challenge, we propose LTRGAD, a two-stage GAD framework that performs feature selection based on local feature-topological residuals (LTR). By processing features separately, LTRGAD effectively introduces topological information while preserving the original local anomalous patterns, enabling more accurate local anomaly detection. Subsequently, global anomaly detection is conducted on the entire graph by leveraging the results from the local detection phase. Extensive experiments on seven benchmark datasets demonstrate the effectiveness of the proposed LTRGAD framework.
Graph Neural Networks (GNNs) have demonstrated strong predictive performance across a wide range of applications. However, their increasing deployment has raised critical fairness concerns, as these models can inherit and amplify existing biases. Most existing fairness approaches rely on explicit demographic information, either directly available or inferred, to measure and mitigate bias. In real-world settings, however, such information is often unavailable or legally prohibited to infer due to privacy concerns, legal restrictions, or regulatory constraints, which substantially limits the applicability of these methods. To address this challenge, we propose Demographic-Independent Fair Graph Learning (DIFGL), a novel framework for fair graph learning without demographic supervision. DIFGL mitigates group unfairness by minimizing disparities in individual treatment across implicitly identified subgroups, thereby enforcing fairness without requiring explicit demographic information. Extensive experiments on benchmark datasets demonstrate that DIFGL achieves significant improvements in fairness while maintaining competitive predictive performance.
Process Reward Models (PRMs) supervise intermediate reasoning steps in large language models (LLMs), but existing PRMs are mainly trained on general-domain data and struggle with the structured, symbolic, and fact-sensitive nature of financial reasoning. Financial tasks require not only correct final answers but also verifiable intermediate steps grounded in domain knowledge. In this paper, we propose Fin-PRM, a domain-specialized, trajectory-aware PRM for financial reasoning that jointly models step-level correctness and trajectory-level coherence, producing binary supervision signals for both local and global reasoning quality. To support reliable supervision, we construct a high-quality financial reasoning dataset of 3K trajectories, where step- and trajectory-level labels are automatically derived from multi-source reward signals, including Monte Carlo rollouts, LLM-based evaluation, and explicit financial knowledge verification. Fin-PRM defines a unified ranking score that integrates step- and trajectory-level rewards, enabling consistent use across multiple settings. We evaluate Fin-PRM in three scenarios: (1) offline trajectory selection for supervised fine-tuning, (2) reward-guided Best-of-N inference for test-time scaling, and (3) process-aware reward shaping for reinforcement learning. Experiments on financial reasoning benchmarks, including CFLUE and FinQA, show that Fin-PRM consistently outperforms general-purpose PRMs and strong baselines. Our project resources will be available at https://github.com/aliyun/qwen-dianjin.
In this work, we propose the first constant-approximation algorithms, LinAst and LinAtg, which simultaneously achieve optimal query complexity O(n) and adaptive complexity O(log n) for non-monotone submodular maximization under a cardinality constraint k over a ground set of size n. Specifically, compared with existing algorithms that attain the best known adaptive complexity of O(log n), our approach preserves this adaptivity while reducing the query complexity from O(n log k) to O(n) and improving the approximation ratio from 0.172-epsilon to 0.193-epsilon. Our algorithms are built upon LinAdapt, which achieves a constant approximation ratio with O(log n) adaptive rounds and linear query complexity by requiring only O(1) candidate guesses of the optimal value. We further introduce the BoostAdapt algorithm, which improves the approximation guarantee to 0.25-\epsilon with O(log n log k) adaptive complexity and O(n log k) query complexity, based on a novel staggered greedy threshold framework that alternately constructs two disjoint solution sets over O(log k) sequential rounds. Extensive experiments on standard benchmark datasets demonstrate that our algorithms consistently outperform state-of-the-art methods in terms of solution quality, query complexity, and running time.
Learning universal representations for time series is fundamental for diverse downstream tasks. However, current approaches largely rely on handcrafted data augmentations, which may distort intrinsic temporal dynamics and structural regularities. In addition, most static representation learning frameworks struggle to cope with the non-stationary nature of real-world time series. To address these issues, we propose Heterogeneous Hypergraph Structure-aware Contrastive Adaptive Network (H²SCAN), a novel augmentation-free framework that derives contrastive supervision directly from graph topology. Specifically, H²SCAN constructs a heterogeneous hypergraph with three node types to capture multi-scale temporal characteristics. Building upon this representation, a meta-adaptation network is introduced to dynamically reweight heterogeneous hyperedges, enabling the model to adapt to distribution shifts in real-time. Finally, a structure-aware contrastive learning objective is employed to align latent representation similarity with the intrinsic hypergraph topology. Experiments on multiple benchmarks and cloud Kafka cluster datasets demonstrate that H²SCAN outperforms existing methods by modeling high-order multi-domain dependencies and preserving the semantic integrity of time series data.
Data selection is a key component of efficient instruction tuning for large language models, as recent work has shown that data quality often matters more than data quantity. Accordingly, prior studies have introduced various multi-dimensional heuristics to evaluate and filter instruction data. However,most existing methods rely on static task-agnostic and model-agnostic weighting schemes, which overlook the varying requirements of specific downstream tasks and the differing pre-existing capabilities of models. In this paper, we propose a framework for learning multi-indicator weights that jointly adapts data selection to both the downstream task and the specific model. Our method identifies optimal weight configurations without full-scale fine-tuning by utilizing in-context learning (ICL) signals on compact tiny-validation sets. These signals serve as efficient performance proxies that ensure high-fidelity evaluation at minimal computational cost. Experiments across multiple benchmarks and model families, including Mistral, Qwen, and Llama, show that the approach achieves performance comparable to or exceeding full-dataset tuning while using only 30% of the training samples on GSM8K. Furthermore, our analysis reveals a trade-off between semantic diversity and logical complexity in reasoning tasks, highlighting the necessity of joint task-model adaptation.
Whole-slide images (WSIs) provide gigapixel-scale visual evidence for cancer diagnosis, yet diagnostically relevant regions are typically sparse and embedded within large amounts of weakly informative tissue. Existing multiple instance learning methods often aggregate all patches by using attention mechanisms or sequence modeling, leading to redundant computation and limited adaptability to slides with highly variable sizes and lesion burdens. We propose Proportional Memory-guided Differential-attention Multiple Instance Learning (PMDMIL) network, a scalable MIL framework that follows a “less is more” principle for WSI classification. PMDMIL retains a fixed ratio of patches, pre-filters noise with a class-wise memory bank, and fuses the sparse survivors through differential attention to amplify subtle yet decisive morphological differences. Experiments on five public WSI benchmarks demonstrate that PMDMIL consistently outperforms representative MIL and sequence-based baselines across multiple evaluation metrics. Notably, on the DHMC dataset, our model can achieve competitive performance, while processing only 10% of patches, indicating that effective instance selection is more important than exhaustive aggregation for WSI classification.
Incomplete multi-view clustering (IMVC) aims to exploit complementary information from multiple views with missing observations. Recent diffusion-based approaches have shown promise for view completion; however, they often fail to capture instance-aligned global correlations across views and suffer from inefficient inference and loosely coupled optimization. In this paper, we propose IASR, an Instance-Aligned Semantic Reconstruction framework for IMVC. IASR formulates missing-view recovery as a cross-view token alignment generator, in which noisy targets, observed views, and timestep embeddings are jointly represented as tokens and interact to capture long-range cross-view dependencies throughout the denoising trajectory. To stabilize representation learning under generative noise, we further introduce a stable–active dual-encoder representation architecture with a generative-adaptive contrastive learning strategy that tightly couples view completion and clustering. Extensive experiments on eight benchmark datasets demonstrate that IASR consistently outperforms state-of-the-art IMVC methods, especially under high missing-rate settings, while achieving improved inference efficiency.
Fairness k-submodular maximization has attracted increasing interest due to its broad relevance in artificial intelligence and machine learning. However, most existing works are limited to monotone objectives or simple size constraints, while non-monotone settings with richer constraints remain largely unexplored. In this paper, we first introduce a constant-factor approximation algorithm for the problem with a general non-monotone objective function under a matroid constraint. Our approach is built upon a two-stage algorithmic framework. Specifically, we first develop an algorithm that guarantees feasibility with respect to upper fairness bounds only. We then show how this algorithm can be systematically extended to simultaneously enforce fairness bounds, while preserving provable approximation guarantees. Comprehensive experiments on standard benchmark datasets demonstrate that our algorithm achieves competitive objective values while maintaining a favorable balance between fairness guarantees and query complexity efficiency compared to existing state-of-the-art methods.