Understanding how humans process visual information during learning remains a central challenge in cognitive neuroscience. Traditional approaches typically rely on behavioral observations or indirect neural measures, which limits direct insight into learners' internal cognitive representations. Recent advances in brain--computer interfaces (BCIs) and deep learning have enabled image reconstruction from electroencephalography (EEG), providing a promising pathway to decode latent cognitive states. In this study, we investigate how EEG-based image reconstruction relates to students' learning. While students watched educational videos, we recorded EEG signals and used an image reconstruction model to generate images reflecting their perceived visual representations. We assessed reconstruction quality using semantic similarity between reconstructed and original images. First, we examine whether semantic similarity predicts overall academic performance and find a strong positive association. Second, we evaluate item-level effects and show that questions answered incorrectly correspond to significantly lower semantic similarity than correctly answered questions. Third, we analyze the temporal dynamics of reconstruction and demonstrate that temporal misalignment patterns---quantified through dispersion-related measures---are associated with learners' performance. Overall, these results suggest that EEG-based image reconstruction captures both the strength and the temporal evolution of learners' cognitive representations, offering a novel tool for studying learning mechanisms and supporting more personalized, data-driven educational interventions.

Examinations hold significant potential for improvement, quantitative investigations of examinations could improve learning efficiency and provide valuable insights for educational decision-making. However, there is a general lack of quantitative methods for investigating examinations. To address this gap, we propose a new metric via complex networks; i.e., the knowledge point network (KPN) of an examination is constructed by representing the knowledge points (concepts, laws, etc.) as nodes and adding links when these points appear in the same question. Then, the topological quantities of KPNs, such as degree, centrality, and community, can be employed to systematically explore the structural properties and evolution of examinations. We analyze 35 physics examinations from the Chinese National College Entrance Examination (2006–2020). We found that the constructed KPNs are scale-free networks that show strong assortativity and small-world effects in most cases. The communities within the KPNs are obvious, and the key nodes are mainly related to mechanics and electromagnetism. Different question types are related to specific knowledge points, leading to noticeable structural variations in KPNs. Moreover, Changes in the KPN topology across years may offer insights guiding college entrance examination reforms. Based on topological quantities such as the average degree, network density, average clustering coefficient, and network transitivity, the comprehensive difficulty coefficient is proposed to evaluate examination difficulty. All the above results show that our approach can objectively and comprehensively quantify the knowledge structures and examination characteristics. These networks may elucidate comprehensive examination knowledge graphs for educators, guiding adjustments in teaching methodologies and supporting data-driven educational decision-making.

This study investigates how emotions co-evolve with interaction structures in online learning communities using Emotional Temporal Networks (ETNs). Fifteen weekly ETNs captured learner-to-learner interactions and node-level emotions across a semester. Regression models related weekly emotions to network properties, and a motif-coloring analysis traced emotional roles within recurrent subgraphs. We also compared network-emotion patterns across learner groups with different achievement levels. Key findings include: (i) Positive emotions predominated but declined over time, while negative and confused emotions increased. The emotion-network association was dynamic and metric-specific, with stable positive links for centrality measures like indegree and eigenvector, but variable or null links for outdegree and betweenness under stress. (ii) Motifs revealed complementary "sharer" (outgoing) and "listener" (incoming) roles: sharers showed greater emotional volatility, whereas listeners maintained more stable positive affect shaped by group context. Two temporal junctures emerged-mid-semester shifts in affect and a pre-exam rise in closed triads and bidirectional ties supporting collaborative problem solving. (iii) High-achieving learners occupied more central, information-rich positions and sustained more stable positive emotions. The ETN framework offers a mesoscale lens for monitoring and guiding emotion dynamics and suggests interventions that enhance information flow, scaffold sharer-listener role dynamics, and use motif-based monitoring for timely instructional support.

Sentiment analysis is crucial in education, yet traditional lexicon-based methods struggle to capture nuanced emotions and group dynamics. We propose a network-based approach using a dataset of 9,873 forum posts from 1,919 MOOC learners. In our sentiment network, learners are nodes in an emotional state space, connected when their emotional distance-defined as the Manhattan distance between their emotion-score vectors-is below a threshold. Analysis reveals a sparse network with emotional variability and small-world characteristics, promoting local clustering and relatively efficient information flow. Learners with lower positive emotions often occupy structurally central positions in the network, characterized by higher degree and betweenness centrality, while high-positive-emotion groups show academic underperformance. Conversely, negative or confused emotions are correlated with better outcomes. Higher network entropy is linked to improved academic performance, indicating enhanced emotional processing. These findings offer new insights into emotional interactions and learning outcomes, suggesting future research on specific emotions and refined emotion analysis in education.

Dialogue datasets are essential for advancing natural language processing (NLP) tasks. However, many existing datasets lack integrated annotations for personality and emotion, limiting models’ ability to effectively capture these aspects and generate personalized, human-like dialogues, which ultimately impact user experience. To address this challenge, we construct bilingual dialogue datasets in Chinese and English, incorporating Big Five personality traits and emotion annotations. We utilize the AutoGen tool within a multi-agent framework to generate multi-turn question-answering dialogue datasets based on fables. By creating persona agents with diverse personalities, we effectively enhance the heterogeneity of personalities, overcoming previous limitations in personality diversity. Finally, we validate the utterance quality in the dataset and investigate the alignment between conversational utterances and speakers’ personality traits. Moreover, by integrating emotional annotations for each utterance, This dataset offers significant potential for developing emotion-aware systems that automatically detect personality traits. It serves as a valuable resource for advancing emotionally intelligent dialogue systems and research in personality and affective computing.

Understanding how humans learn remains a considerable challenge in the field of education. This study investigates the knowledge acquisition process in online learning from an ecological perspective, using 77,301 behavioral data from an online platform. Employing ecological network analysis, the study constructs attention flow networks to visualize and compare the learning behaviors of 94 dropouts and 287 nondropouts. Key findings highlight structural differences in attention flow networks, emphasizing the focused learning behaviors among nondropouts. Additionally, hierarchical trophic structure diagrams illustrate the organization of learning resources, with nondropouts demonstrating a more coherent structure conducive to knowledge acquisition. The study underscores the critical role of designing course information and courseware resources. It advocates optimizing these resources as primary entry points to initiate and sustain knowledge acquisition. Furthermore, the importance of clear learning goals is highlighted, emphasizing their role in guiding learners towards persistent and effective learning. Timely testing and review of course content are also recommended to consolidate knowledge and reduce distraction, particularly through the facilitation of problem-solving activities and feedback mechanisms. Symbolic regression analysis contributes to developing interpretable models that identify pivotal factors influencing sustained learning, such as the design of test resources and support mechanisms for problem-solving. Finally, stability analysis reveals that the nondropout network exhibits superior stability. These insights provide practical guidelines for enhancing e-learning course design, ultimately promoting continuous engagement and mitigating dropout rates in online learning environments.

Mining learner preferences and needs from individual learning behavior data is a critical task in course recommendation systems. While graph-based models have shown efficacy in capturing pairwise relationships between learners and courses, they often overlook the complex higher-order interactions involving learners, courses and teachers that are essential for accurate recommendations. To address this limitation, we propose a novel Hypergraph Convolutional Network for Course Recommendation (HCNCR) framework, designed to model these higher-order interactions effectively. Our approach constructs course and learner hypergraphs based on course attributes and learner similarity relations, respectively. By employing hypergraph convolution, we capture the intrinsic higher-order relationships within these hypergraphs. Additionally, we utilize graph convolutional layers on the learner-course bipartite graph to integrate embeddings derived from hypergraphs, achieving comprehensive representations of both learners and courses. Extensive experiments conducted on real-world datasets demonstrate that HCNCR significantly outperforms existing state-of-the-art methods in course recommendation tasks.

Identifying the epistemic emotions of learner-generated reviews in massive open online courses (MOOCs) can help instructors provide adaptive guidance and interventions for learners. The epistemic emotion identification task is a fine-grained identification task that contains multiple categories of emotions arising during the learning process. Previous studies only consider emotional or semantic information within the review texts alone, which leads to insufficient feature representation. In addition, some categories of epistemic emotions are ambiguously distributed in feature space, making them hard to be distinguished. In this article, we present an emotion-semantic-aware dual contrastive learning (ES-DCL) approach to tackle these issues. In order to learn sufficient feature representation, implicit semantic features and human-interpretable emotional features are, respectively, extracted from two different views to form complementary emotional-semantic features. On this basis, by leveraging the experience of domain experts and the input emotional-semantic features, two types of contrastive losses (label contrastive loss and feature contrastive loss) are formulated. They are designed to train the discriminative distribution of emotional-semantic features in the sample space and to solve the anisotropy problem between different categories of epistemic emotions. The proposed ES-DCL is compared with 11 other baseline models on four different disciplinary MOOCs review datasets. Extensive experimental results show that our approach improves the performance of epistemic emotion identification, and significantly outperforms state-of-the-art deep learning-based methods in learning more discriminative sentence representations.

Blended learning, as an efficient teaching mode that combines the advantages of both online and offline learning, has been widely applied in universities. Nevertheless, the different learning patterns induce difficulty in evaluating the learning quality. In this paper, an approach of integrating online and offline interactions is proposed by constructing a weighted multiplex network (WMN), in which online communication behavior and offline peer relations are represented as edges in respective network layers, and edge weight depends on the frequency of interactions. Under the framework of WMNs, learners’ attributions such as behavior, sentiment and cognition can be systematically analyzed. We use a case study to compare the differences in various indicators between the online and offline networks, and investigate the relationships between network structure and individual sentiment, cognition and grade, respectively. Results show that the correlations between network centrality and cognition or grade are significantly improved in the WMN, which demonstrate WMNs have natural advantages in the analysis of blended learning. This study provides methodological and practical implications for the analysis and understanding of learner multiple interactions, which might contribute to improving the dynamic regulation and accurate guidance of blended learning processes and optimizing existing teaching models.

Understanding the relationship between interactive behaviours and discourse content has critical implications for instructors' design and facilitation of collaborative discussion activities in the online discussion forum (ODF). This paper adopts social network analysis (SNA) and epistemic network analysis (ENA) methods to jointly investigate the relationships between students’ network characteristics, discussion topics, and learning outcomes in a course discussion forum. Discourse data from 207 participants were included in this study. The findings indicated that (1) the interactive network generated in the collaborative discussion activities was sparsely connected, and there was limited information exchange between instructors and students; (2) students’ discussion topics were mainly related to the learning content; (3) compared with the isolated group, students in the leader, mediator, and animator groups were more concerned about topics related to the learning content; and (4) students who discussed more topics related to the learning content performed better than the students who discussed more topics related to learning methods and social interactions. The learning outcomes of the influencer and leader groups were significantly higher than those of the peripheral and isolated groups. However, there was no significant correlation between students’ individual centrality and their learning outcomes. The findings enrich the ODF research on the comprehensive identification of interactive behaviours and discourse content in the process of collaborative discussion activities and on the discussion topic differences between different role groups. The study findings also have practical implications for instructors to design effective instructional interventions aimed at improving the quality of collaboration in the ODF.

This paper focuses on the analysis of textbooks using complex networks theory. The structure features of subject knowledge networks are examined by extracting knowledge points from the text and representing them as nodes, while the co-occurrence relationships between the knowledge points are represented as edges. Gephi and Networkx are utilized to visualize and analyze the network topology characteristics, providing objective technical support for textbook analysis. As a case study, this paper selects Chinese high school information technology subject textbooks to construct subject knowledge networks. Through a comparison and comprehensive analysis of different versions of textbooks networks, this study uncovers the differences and characteristics of various textbook versions, as well as the deep structure and internal logic of textbooks. The results demonstrate that textbook analysis based on complex networks offers many advantages, including interpretability, expressiveness, generalization, and flexibility.

Online learning community provides abundant semantic text information for investigating learning behaviors. Despite the increasing of learning behavior in higher education, few previous researches have studied online learning from multidimensional behavior. This study mainly explores the interrelationship of cognition, emotion and interaction when learners engage in online discussion. The research object is online discussion textual from a certain course. First, lag sequence analysis is adopted to analyze the dynamic of cognition from the micro perspective, and the relationship between cognition and emotion combined with emotion analysis is investigated. Furthermore, this study proposes the standards of quantifying cognition and emotion from the macro perspective, and the cognition, emotion and interaction are analyzed in a unified framework by using social network analysis. Our findings suggest that: (1) Cognitive level is stable during the learning process, and can be improved by continuous thinking and analysis. (2) Positive emotion plays a significant role in developing higher-level cognition, and its value increases gradually with the improvement of cognition level. (3) Network structure has important influence on individual cognition, who with higher cognitive levels are usually in the center of the network and have larger interaction quality. (4) The learning community with more intensive interactions show higher positive emotion, indicating that emotion transmission is realized by means of network structure. This study might give theoretical and technical supports for helping learners improve the learning quality and efficiency in the online learning.

Studying the networked nature of social and cognitive aspects of learner interactions is the key to understanding how successful collaborative learning occurs in asynchronous online discussion forums (AODFs). Guided by network science and multiplex network analysis, this study compared the differences of network structure and properties between the social (learners as nodes and commenting on the others' contributions as edges) and cognitive (learners as nodes and explicitly quoting the others' contributions as edges) networks within an AODF. It additionally examined the differences of individual measures between these two monolayer networks and their integrated two multiplex networks. The two multiplex networks were respectively framed in superimposed and unfold approaches. Moreover, correlation analysis was conducted to examine the relationships between individual measures and learning performance across the four networks. Results showed key differences at the network and individual levels between the social and cognitive networks. The two networks had different compositions of participants, and students in the cognitive network occupied more central positions in general. Besides, certain individual measures in the multiplex networks were relatively higher and more related to learning performance than those in the monolayer networks. These results indicate that integrating multi-aspect information of collaboration may be more conducive to unveiling learners' interaction patterns in asynchronous online discussions. Practitioner notes What is already known about this topic? Studying the networked nature of the social and cognitive aspects of learner interactions is the key to understanding how successful collaborative learning occurs. Social network analysis and content analysis are two commonly used methods to analyse learner interactions in asynchronous online discussions. There is a lack of research examining learners' overall interaction patterns by integrating social and cognitive interactions. What this paper adds? This study framed cognitive interactions as a networked phenomenon to unveil the way learners participate in the cognitive aspect of learner interactions. Multiplex network analysis (MNA) was introduced to integrate social and cognitive interactions to understand learners' overall interaction patterns. Integrating social and cognitive interactions led to a more closer relationship between learner interactions and learning performance. Implications for practice and/or policy? The proposed approach for framing the cognitive network and the introduced approach of MNA could be jointly employed to discern learners' overall interaction pattern in asynchronous online discussions. Course facilitators and instructors should be aware of the inconsistencies in social and cognitive interactions and then simultaneously examine them so as to acquire an accurate understanding of learners' diverse interactions. Pedagogical strategies for simultaneously facilitating social and cognitive interactions could be used to better improve learning performance.

利用新型分析方法综合分析学习者交互以获取关于其学习过程的有益见解,是在线教育研究和实践的迫切需求.文章以中部某高校的一门科教融合创新在线课程为例,采用社会网络分析和多层网络分析方法,分别对学习者的在线同步和异步交互模式进行比较与整合分析,并探讨其与学习成绩之间的关系.研究结果发现:相比同步交互网络,异步交互网络更为稠密,且其个体指标值相对更高;相比同步和异步两个单层交互网络,基于二者整合而得到的多层交互网络在网络连边数量上显著增多,在个体指标值上显著增高,而且学习者在多层和单层交互网络中的地位相对不同;相比单层交互网络,多层交互网络的部分个体指标与学习成绩的相关性相对更高.研究为分析和理解学习者交互提供了研究方法论和教学实践方面的启示性意义,有助于改进对在线群体学习过程的动态调控与精确引导并优化现有教学模式.

In the field of learning analytics, mining the regularities of social interaction and cognitive processing have drawn increasing attention. Nevertheless, in MOOCs, there is a lack of investigations on the combination of social and cognitive behavioral patterns. To fill in this gap, this study aimed to uncover the relationship between social interaction, cognitive processing, and learning achievements in a MOOC discussion forum. Specifically, we collected the 3925 participants’ forum data throughout 16 weeks. Social network analysis and epistemic network analysis were jointly adopted to investigate differences in social interaction, cognitive processing between two achievement groups, and the differences in cognitive processing networks between two types of communities. Finally, moderation analysis was employed to examine the moderating effect of community types between cognitive processing and learning achievements. Results indicated that: (1) the high- and low-achieving groups presented significant differences in terms of degree, betweenness, and eigenvector centrality; (2) the stronger cognitive connections were found within the high-achieving group and the instructor-led community; (3) the cognitive processing indicators including insight, discrepancy, and tentative were significantly negative predictors of learning achievements, whereas inhibition and exclusive were significantly positive predictors; (4) the community type moderated the relationship between cognitive processing and learning achievements.

Emotional design in multimedia learning is an instructional design method, which can induce positive emotional experiences for students to improve their intrinsic motivation, thus facilitating their learning outcomes. However, previous studies have rarely considered the impact of positive emotional design on learning outcomes when students were in stressful situations. In this regard, this study explored the role of positive emotional design principles in students' stress-arousal state. Fifty students from a university in central China were randomly divided into two groups, 25 as the experimental group and 25 as the control group. On the premise of completing the stress arousal task, two groups studied the multimedia courses of positive emotion design and neutral emotion design respectively by using ECG equipment and questionnaires. The results of the study show that students can learn materials with positive emotion design to effectively prevent the decline of transfer performance caused by the stressful situation.

Investigating behavioral characteristics of college students is of great importance to grasp students' behaviors and assist administrators make efficient manage strategies. In this paper, our research focuses on the discovery of burst nature and memory effect of selecting behavior and repeating behavior, which are universal characteristics of human beings, from the campus big data. To this end, we firstly analysis the interevent time distributions of selecting canteens and entering the library, respectively. Results show that both of these behaviors significantly follow heavy-tail power-law distribution, which verify the preference of these behaviors. Then, burstiness parameter B and memory M are measured based on interevent time series, and we observe that the corresponding distributions approximately follow Gaussian forms. Furthermore, we analyze the relations between students' GPA and B or M. The main results indicate that GPA not only decays with the value of burst of entering the library, but also is directly proportional to the absolute value of memory of selecting canteens.

人工智能技术变革下的机器辅助教学为教育教学发展注入新的动能.机器辅助教学的内涵与外延在被人工智能技术重塑的同时,其结构也在智慧性、主体性和协同性等方面有新发展.然而,随着教学机器服务能力不断增强,范畴不断拓展,如何对其教学成效展开合理评测成为亟待解决的重要议题.机器辅助教学的能力向度是指机器赋能教学过程的指向目标与能力层级,对其剖析并构建能力向度模型,能以全新视角全面认知教学机器的客观概况与内部逻辑,又可从智能教学机器应用中提取可观测、可测量、可解释的关键因素,借以解决能力评测问题.本研究基于对机器辅助教学基本内涵的探讨,确定机器辅助教学的价值取向、系统框架与关键要素,构建并验证了以教育主体、场景、资源和数据为承载的机器辅助教学能力向度模型.这一模型研究需要秉持教育伦理先导的理念,坚持数据与理论双向驱动,推动人机协同并进,结合因果推断增强数据解释,从而为机器辅助教学能力评测标准的制定与实践提供理论参照和方向指引.

Analyzing and mining students' behaviors and interactions from big data is an essential part of education data mining. Based on the data of campus smart cards, which include not only static demographic information but also dynamic behavioral data from more than 30000 anonymous students, in this paper, the evolution features of friendship and the relations between behavior characters and student interactions are investigated. On the one hand, four different evolving friendship networks are constructed by means of the friend ties proposed in this paper, which are extracted from monthly consumption records. In addition, the features of the giant connected components (GCCs) of friendship networks are analyzed via social network analysis (SNA) and percolation theory. On the other hand, two high-level behavior characters, orderliness and diligence, are adopted to analyze their associations with student interactions. Our experiment/empirical results indicate that the sizes of friendship networks have declined with time growth and both the small-world effect and power-law degree distribution are found in friendship networks. Second, the results of the assortativity coefficient of both orderliness and diligence verify that there are strong peer effects among students. Finally, the percolation analysis of orderliness on friendship networks shows that a phase transition exists, which is enlightening in that swarm intelligence can be realized by intervening the key students near the transition point.

情绪与社会化交互是构建学习者模型的两个重要特征,学习者的情绪特征计算与社会网络分析已获得了学习分析领域的广泛关注.本研究以某高校云平台上两门课程的论坛发帖纪录为研究对象,分别探究了学习群体在情绪表征(积极、消极与困惑情绪密度值)以及社会网络交互(网络中心性与网络结构特征)方面的差异.研究结果表明:具有网络交互关系的学习者间情绪倾向趋于一致.整体网络中的"子团体"以分散型网络居多,不同社区组学习者在积极和消极情绪上具有显著性差异,而在困惑情绪上差异性并不显著.与中、低成效组相比,高成效组学习者在信息传递的中介性以及协作学习中的参与性方面的表现更为显著,但其积极情绪密度较低.论坛中情绪和交互特征的联合分析有助于对学业风险个体的准确干预和群体互动质量的提升.