Teachers' emotional states are critical in educational scenarios, profoundly impacting teaching efficacy, student engagement, and learning achievements. However, existing studies often fail to accurately capture teachers' emotions due to the performative nature and overlook the critical impact of instructional information on emotional expression.In this paper, we systematically investigate teacher sentiment analysis by building both the dataset and the model accordingly. We construct the first large-scale teacher multimodal sentiment analysis dataset, T-MED.To ensure labeling accuracy and efficiency, we employ a human-machine collaborative labeling process.The T-MED dataset includes 14,938 instances of teacher emotional data from 250 real classrooms across 11 subjects ranging from K-12 to higher education, integrating multimodal text, audio, video, and instructional information.Furthermore, we propose a novel asymmetric attention-based multimodal teacher sentiment analysis model, AAM-TSA.AAM-TSA introduces an asymmetric attention mechanism and hierarchical gating unit to enable differentiated cross-modal feature fusion and precise emotional classification. Experimental results demonstrate that AAM-TSA significantly outperforms existing state-of-the-art methods in terms of accuracy and interpretability on the T-MED dataset.

Vision-language models (VLMs) have gained widespread attention for their strong zero-shot capabilities across numerous downstream tasks. However, these models assume that each test image’s class label is drawn from a predefined label set and lack a reliable mechanism to reject samples from emerging unknown classes when only unlabeled data are available. To address this gap, open-set domain adaptation methods retrain models to push potential unknowns away from known clusters. Yet, some unknown samples remain stably anchored to specific known classes in the VLM feature space due to semantic relevance, which is termed as Semantic Affinity Anchoring (SAA). Forcibly repelling these samples unavoidably distorts the native geometry of VLMs and degrades performance. Meanwhile, existing score‑based unknown detectors use simplistic thresholds and suffer from threshold sensitivity, resulting in sub‑optimal performance. To address aforementioned issues, we propose VLM-OpenXpert, which comprises two training‑free, plug‑and‑play inference modules. SUFF performs SVD on high-confidence unknowns to extract a low-rank "unknown subspace". Each sample’s projection onto this subspace is weighted and softly removed from its feature, suppressing unknown components while preserving semantics. BGAT corrects score skewness via a Box–Cox transform, then fits a bimodal Gaussian mixture to adaptively estimate the optimal threshold balancing known-class recognition and unknown-class rejection. Experiments on 9 benchmarks and three backbones (CLIP, SigLIP, ALIGN) under Source-Free OSDA settings show that our training-free pipeline matches or outperforms retraining-heavy state-of-the-art methods, establishing a powerful lightweight inference calibration paradigm for open-set VLM deployment.

Publisher Copyright: © 2026 Owner/Author.

Graph neural networks have recently demonstrated remarkable performance in predicting material properties. Crystalline material data is manually encoded into graph representations. Existing methods incorporate different attributes into constructing representations to satisfy the constraints arising from symmetries of material structure. However, existing methods for obtaining graph representations are specific to certain constraints, which are ineffective when facing new constraints. In this work, we propose a code generation framework with multiple large language model agents to obtain representations named Rep-CodeGen with three iterative stages simulating an evolutionary algorithm. To the best of our knowledge, Rep-CodeGen is the first framework for automatically generating code to obtain representations that can be used when facing new constraints. Furthermore, a type of representation from generated codes by our framework satisfies six constraints, with codes satisfying three constraints as bases. Extensive experiments on two real-world material datasets show that a property prediction method based on such a graph representation achieves state-of-the-art performance in material property prediction tasks.

Graph anomaly detection methods based on contrastive learning have been widely proposed due to their powerful unsupervised feature learning capabilities. However, most existing methods overlook the loss and degradation of original graph information caused by the random data augmentation. In addition, the impact of anomalous nodes is usually ignored during the training phase. To address these issues, we propose a collaborative contrastive self-supervised graph anomaly detection model, namely, CCS-GAD. The core idea lies in the collaboration and mutual constraints among its submodules. Specifically, we design a view collaboration module to seamlessly integrate contrastive-based and reconstruction-based frameworks, capturing node patterns from both local and global views simultaneously to alleviate the graph information gaps caused by data augmentation in conventional self-supervised paradigms. Additionally, we introduce a filtering collaboration method to identify and filter out potential anomalies exhibiting abnormal behaviors from other nodes, thereby reducing their interference with the model’s learning of normal patterns. Finally, anomalies are detected based on local consistency and global reconstruction errors. Extensive experiments on six real-world datasets with seven state-of-the-art(SOTA) baselines demonstrate the superior adaptability of our model in graph anomaly detection. Specifically, our model achieves an average improvement of 16.43% in AUC, 5.83% in Precision, and 9.48% in F1-score.

Property prediction is a fundamental task in crystal material research. To model atoms and structures, structures represented as graphs are widely used and graph learning-based methods have achieved significant progress. Bond angles and bond distances are two key structural information that greatly influence crystal properties. However, most of the existing works only consider bond distances and overlook bond angles. The main challenge lies in the time cost of handling bond angles caused by the large cutoff, which leads to a significant increase in inference time. To address this issue, we first propose a flexible dual-scale neighbor partitioning mechanism with consistent receptive fields in point graphs and edge graphs, which reduces redundant information unrelated to modeling periodicity and improves prediction speed without affecting the modeling of graph periodicity. Then, we propose a novel Atom-Distance-Angle Graph Neural Network (ADA-GNN) for property prediction tasks, which processes node information and structural information separately. The accuracy of predictions and the speed of inference process are both improved with the flexible dual-scale modeling and the specially designed architecture of ADA-GNN. The experimental results validate that our approach achieves state-of-the-art results in two large-scale material benchmark datasets on property prediction tasks.

The cooperation in mixed cooperative-competitive tasks has drawn significant attention in multi-agent deep reinforcement learning. Agents need to cooperate with their teammates while competing against their opponents. However, most existing works treat the cooperative agents and competitive agents equally as they perform the same operation on all the agents. As a result, without distinguishing between cooperative and competitive agents, they may suffer from information disorder in learning an optimally cooperative policy and struggle to decide on the next step action. To address the above issues, we decompose the final Q-value into a weighted combination of three parts: the Q-values of the cooperative agents, the competitive group, and the current agent. A theoretical proof of the correctness of the decomposition is provided. With this decomposition, we are able to consider cooperative and competitive agents separately. Accordingly, we propose a multi-agent actor-critic algorithm called actor-double-attention-critic (ADAC) under centralized training and decentralized execution according to the decomposition. In ADAC, networks with group-specific attention and an attentional weighting network are specially designed. With the designed double-attention structure, ADAC can capture the distributions from different agents and improve cooperation performance. Extensive experiments are conducted in three scenarios with nine settings against six representative methods. The results demonstrate the superiority of the proposed ADAC model against state-of-the-art methods in various mixed cooperative-competitive tasks. The code is available at https://github.com/CrazyBayes/ADAC

Fault Scenario Identification (FSI) is a challenging task that aims to automatically identify the fault types in communication networks from massive alarms to guarantee effective fault recoveries. Existing methods are developed based on rules, which are not accurate enough due to the mismatching issue. In this paper, we propose an effective method named Knowledge-Enhanced Graph Neural Network (KE-GNN), the main idea of which is to integrate the advantages of both the rules and GNN. This work is the first work that employs GNN and rules to tackle the FSI task. Specifically, we encode knowledge using propositional logic and map them into a knowledge space. Then, we elaborately design a teacher-student scheme to minimize the distance between the knowledge embedding and the prediction of GNN, integrating knowledge and enhancing the GNN. To validate the performance of the proposed method, we collected and labeled three real-world 5G fault scenario datasets. Extensive evaluation conducted on these datasets indicates that our method achieves the best performance compared with other representative methods, improving the accuracy by up to 8.10%. Furthermore, the proposed method achieves the best performance against a small dataset setting and can be effectively applied to a new carrier site with a different topology structure.

Trust prediction provides valuable support for decision making, information dissemination, and product promotion in online social networks. As a complex concept in the social network community, trust relationships among people can be established virtually based on: 1) their interaction behaviors, e.g., the ratings and comments that they provided; 2) the contextual information associated with their interactions, e.g., location and culture; and 3) the relative temporal features of interactions and the time periods when the trust relationships hold. Most of the existing works only focus on some aspects of trust, and there is not a comprehensive study of user trust development that considers and incorporates 1)-3) in trust prediction. In this article, we propose a context-aware deep trust prediction model C-DeepTrust to fill this gap. First, we conduct user feature modeling to obtain the user's static and dynamic preference features in each context. Static user preference features are obtained from all the ratings and reviews that a user provided, while dynamic user preference features are obtained from the items rated/reviewed by the user in time series. The obtained context-aware user features are then combined and fed into the multilayer projection structure to further mine the context-aware latent features. Finally, the context-aware trust relationships between users are calculated by their context-aware feature vector cosine similarities according to the social homophily theory, which shows a pervasive property of social networks that trust relationships are more likely to be developed among similar people. Extensive experiments conducted on two real-world datasets show the superior performance of our approach compared with the representative baseline methods.

A redundancy-based strategy is widely employed by assigning each task to multiple workers and then inferring the correct answer (called truth) for each task in crowdsourcing. Most existing truth inference methods are designed for the situation with a fairly big number of answers for each task (referred to as high redundancy). However, the high redundancy unavoidably leads to a high cost. In this work, we propose a novel truth inference approach called TWLR based on worker representations for the situation with a small number of answers for each task (referred to as low redundancy). We develop a deep model to learn the representations of workers considering both answers and worker-task relations. For each task, we identify the worker with the highest quality, and select his/her answer as the predicted answer. To the best of our knowledge, this is the first work to perform truth inference by utilizing deep learning techniques to deal with the low redundancy situation in crowdsourcing. We have conducted a set of experiments against 7 real-world datasets to show the accuracy improvement of our truth inference approach by comparing with 11 baseline methods.

Crowdsourcing platforms are widely used for requesters to find workers for general tasks. The answers to general tasks are usually open and not constrained by multiple choices. For the general tasks, the worker performance prediction models can facilitate the task assignment process in crowdsourcing. Worker performance prediction is affected by the three roles: the worker, the requester, and the task. The existing worker performance prediction models mainly consider the features of tasks and workers. However, these models rarely consider the features of requesters. And the existing worker performance prediction models for multiple-choice tasks are not suitable for general tasks as they are built based on the workers' accuracy on choices. In this work, we propose a worker performance prediction model by taking account of features of workers, tasks, and requesters to help requesters select workers for their general tasks on crowdsourcing platforms. We design a relationship learning module to learn the low dimension relationship representations of workers, tasks, and requesters. Furthermore, we design a performance learning model to predict workers' performance based on the features and relationship representations of workers, tasks, and requesters. A set of experiments against the realworld dataset from the Zhubajie platform has been conducted. Experimental results show that the proposed approach has better prediction results than the existing baseline methods.

Trust relationship prediction among people provides valuable supports for decision making, information dissemination, and product promotion in online social networks. Network embedding has achieved promising performance for link prediction by learning node representations that encode intrinsic network structures. However, most of the existing network embedding solutions cannot effectively capture the properties of a trust network that has directed edges and nodes with in/out links. Furthermore, there usually exist rich user attributes in trust networks, such as ratings, reviews, and the rated/reviewed items, which may exert significant impacts on the formation of trust relationships. It is still lacking a network embedding-based method that can adequately integrate these properties for trust prediction. In this work, we develop an AtNE-Trust model to address these issues. We firstly capture user embedding from both the trust network structures and user attributes. Then we design a deep multi-view representation learning module to further mine and fuse the obtained user embedding. Finally, a trust evaluation module is developed to predict the trust relationships between users. Representation learning and trust evaluation are optimized together to capture high-quality user embedding and make accurate predictions simultaneously. A set of experiments against the real-world datasets demonstrates the effectiveness of the proposed approach.

General tasks on crowdsourcing platforms attract more and more workers with different skills and experiences. Existing approaches only leverage the information from tasks with feedback to evaluate worker ability. However, there are millions of tasks without feedback on the platforms. The participating behavior of workers involved in these tasks has not been exploited. In this work, we propose a worker ability model PB-Worker to support general tasks on crowdsourcing platforms. We model the worker latent relation and task latent relation by exploiting the worker participating behavior. To the best of our knowledge, this is the first work to consider the worker participating behavior. Our model is a semi-supervised model that can cover tasks with feedback and tasks without feedback. We employ the ladder network to generate the representations of workers and employ the neural network to predict the worker ability scores. A set of experiments against the real-world dataset from the Zhubajie platform has been conducted. Experimental results show that the output quality of the proposed approach is better than the existing baseline methods.

Trust prediction in online social networks is crucial for information dissemination, product promotion, and decision making. Existing work on trust prediction mainly utilizes the network structure or the low-rank approximation of a trust network. These approaches can suffer from the problem of data sparsity and prediction accuracy. Inspired by the homophily theory, which shows a pervasive feature of social and economic networks that trust relations tend to be developed among similar people, we propose a novel deep user model for trust prediction based on user similarity measurement. It is a comprehensive data sparsity insensitive model that combines a user review behavior and the item characteristics that this user is interested in. With this user model, we firstly generate a user's latent features mined from user review behavior and the item properties that the user cares. Then we develop a pair-wise deep neural network to further learn and represent these user features. Finally, we measure the trust relations between a pair of people by calculating the user feature vector cosine similarity. Extensive experiments are conducted on two real-world datasets, which demonstrate the superior performance of the proposed approach over the representative baseline works.