The quality of social interaction in online courses often falls short of expectations, probably due to learners' difficulties in perceiving the sociality of online learning environments and accessing valid information about others. Group awareness tools (GATs) can address this issue; however, there is little research on GATs specifically designed for online learning communities. Furthermore, different types of awareness are typically provided without meaningful integration. This study introduced an aggregated group awareness tool named SKN in the form of a social knowledge network, which presents a systematic visualization of behavioral, social, and cognitive information about learners within the online community. A quasi-experimental design was employed to investigate SKN's impact on learners' perceptions of the online learning environment and social interactions. The experimental group (n = 21) utilized SKN, while the control group (n = 20) did not have access. Results revealed that the experimental group demonstrated significantly enhanced social perceptions of the online environment and engaged in more social interactions compared to the control group. Interviews indicated high levels of satisfaction with the SKN tool, confirming its positive influence on the online learning experience. These findings underscore the potential of aggregated group awareness approaches for enhancing the social attributes of the online learning environment.

The emergence of generative AI has profoundly impacted the field of education by enabling individuals to enhance both the efficiency and quality of cognitive tasks by delegating part of the tasks to generative AI.This process is referred to as cognitive outsourcing.However,individuals' effectiveness in using AI varies.Empirical research on the educational applications of generative AI remains limited,primarily focusing on evaluating technical capabilities and the effects of learning support.At present,the cognitive and behavioral prerequisites for effective cognitive outsourcing remain unclear.Furthermore,the differences in prior knowledge,behavioral patterns,and cognitive structures among individuals with varying performances have yet to be thoroughly explored. In this study,we designed a cognitive outsourcing activity for graduate students involving a sample of 46 participants(10 males,36 females;age:M=26.39,SD=6.91).The activity consisted of two sessions.In the first session,participants were allotted 30 minutes to independently construct a concept map on the topic"Artificial Intelligence and Teachers"using pen and paper,which served as a measure of their prior knowledge.In the second session,participants engaged with a generative AI system to compose an essay on the same topic within a 100-minute time frame using a computer.The entire process was video-recorded.Based on expert evaluations,participants were categorized into high-performance and low-performance groups according to their essay scores.Interactive behaviors and contents were coded,and behavioral sequence transitions between the two groups were mapped using Lag Sequence Analysis.Additionally,Epistemic Network Analysis was employed to construct cognitive structure mappings,followed by a comparative analysis of the differences between the two groups. The results indicate that the high-performance group exhibited significantly higher prior domain knowledge compared to the low-performance group.Significant differences were observed between the two groups,including the frequency of different interactive behaviors,the frequency of different cognitive elements,the behavioral sequences,and the cognitive network structures.From the behavioral perspective,the high-performance group demonstrated significantly more diversified behavioral transitions,forming a distinctive pattern characterized by"rapid and autonomous task comprehension and planning,efficient and precise human-computer interaction,selective information extraction and deep processing."From the cognitive perspective,the high-performance group exhibited a well-balanced and comprehensive cognitive structure characterized by diverse and tightly interconnected cognitive elements.In contrast,the low-performance group displayed an unbalanced and loosely connected cognitive structure,primarily engaging with lower cognitive-level interaction.Overall,the findings indicate that effective cognitive outsourcing is a multifaceted process that necessitates active participation and profound cognitive processing.It demands proficient integration between internal cognitive frameworks and external technological tools. These findings highlight the distinct behavioral patterns and cognitive structures of individuals with varying levels of success in cognitive outsourcing activities and elucidate the cognitive and behavioral requirements for effective cognitive outsourcing.By focusing on individuals' prior knowledge and interactive processes,this study examines the influence of cognitive and behavioral characteristics on the efficacy of generative AI-assisted writing,thus contributing to empirical research on generative AI-supported education.Additionally,it extends the theoretical understanding of cognitive outsourcing and provides insight for future research and educational practices.Furthermore,the interactive behavior and content coding framework established in this study,along with the application of Lag Sequence Analysis and Epistemic Network Analysis,provide valuable methodological references.Future research should further investigate the long-term and deep-seated effects of cognitive outsourcing on individuals with different characteristics,as well as the intrinsic neural mechanisms underlying effective cognitive outsourcing.

Cross-prompt trait scoring task aims to learn generalizable scoring capabilities from source- prompt data, enabling automatic scoring across multiple dimensions on unseen essays. Existing research on cross-prompt trait essay scoring primarily focuses on improving model generalization by obtaining prompt-invariant representations. In this paper, we approach the research problem from a different perspective on invariance learning and propose a scoring-invariant learning objective. This objective encourages the model to focus on intrinsic information within the essay that reflects its quality during training, thereby learning generic scoring features. To further enhance the model’s ability to score across multiple dimensions, we introduce a trait feature extraction network based on routing gates into the scoring architecture and propose a trait consistency scoring objective to encourage the model to balance the diversity of trait-specific features with scoring consistency across traits when learning trait-specific essay features. Extensive experiments demonstrate the effectiveness of our approach, showing advantages in multi-trait scoring performance and achieving significant improvements with low-resource prompts.

As the core element of educational digital transformation, data holds significant value in empowering the enhancement of educational quality. However, current applications of educational data at both institutional and regional levels has not reached their expected potentials. The primary issue lies in that data has not been actively embedded in the diverse and complex educational processes as a service to facilitate the solution of practical problems. In order to solve the above problems, this paper proposed an u201Cembeddedu201D education data service ecosystem, and discussed its construction process from three aspects: first, realizing the change of application-oriented data concept to grasp the relationship between data service and education governance from the new concept; second, designing the technical path of an u201Cembeddedu201D data service ecosystem, infiltrating and transforming the original business through u201Cembeddedu201D data service to realize the application scenario transformation of u201Cdata searchu201D; third is to establish the operation guarantee of the u201Cembeddedu201D education data service ecology to ensure the stable operation of the data service ecology and the continuous empowerment of regional education governance. Through research, this paper was expected to provide a systematic solution for regions and schools to construct a data-enabled education service ecosystem.

Learning resources are quite important for online learning while resource provision based on algorithms could not address learners' ubiquitous needs well. Moreover, the structure and content of resources are pre-defined which makes the "Structure" and "Content" coupled closely and could not easily adjust when learners' needs changed. To solve this problem, an automatic resource generation mechanism is needed. In this study, we summarize the main components of resource design and proposed a "Structure-Content Loosely Coupled" resource model (Learning Cell Model). The model separates the structure and content into independent yet connected parts by defining "Dynamic Structure" and "Container". Then, the automatic resource generation mechanism and its supporting system were designed based on the model and used in two 5th Grade classes. Results showed the mechanism and system could generate resources according to learners' needs accurately and improve learners' learning outcomes without increasing their cognitive load. Further, the learners had good attitude, technique acceptance, and satisfaction. Overall, the "Structure-Content Loosely Coupled" model and the proposed mechanism could be used creatively for more flexible and adaptive resource provision. They made the resource generation timely and automatic which helped teachers' resource design. The results are enlightening and foster further research in this field.

Chinese as a second language (CSL) learning has attracted more attention and supporting learners with adaptive resources becomes difficult. Some online systems recommended pre-designed resources from existing databases while the resources could not match learners' context. Designing resources dynamically according to learners' needs could be a solution while it's time-consuming. Targeting this problem, we proposed a "content-structure" loosely coupled model. Based on the model, we developed an automatic resource generation system and used it in a university. One class was chosen and the students were randomly assigned to the experimental (22 students) and control group (21 students). They participated in the course all the same except the resources generation method. During the learning process, the online behaviors were recorded for behavioral analysis and the learners' learning outcome and perceptions were measured by tests and questionnaires. Results showed that the system played positive roles in improving learners' learning outcome and perceptions. Moreover, we found that learners in experimental group participated more actively and there's evidence that the system could help learners better reflect on their needs. The results revealed the effectiveness of the system in supporting CSL learners' contextualized learning. This design will provide inspiration for future context-aware CSL learning research.

Information and communication technologies (ICTs) in basic education, i.e. K12 education, have often been constrained by sociocultural context and limited to pedagogical aspects, and these innovations have not yet catalyzed curriculum reform as expected. To address this research gap, this study adopted a key-competence-based subject knowledge learning tool, i.e. Smart Learning Partner (SLP), and conducted a quasi-experimental study to evaluate its effects among middle school students. A total of 400 participants in Grade 7 from four schools were followed for a period of two and a half years. The results on online learning engagement indicated that the manipulation generally succeeded in promoting students' use of SLP in the experimental schools. Furthermore, the results on academic performance showed that SLP could not only directly benefit students' learning in the first academic year but also continued to have an effect within the following one and a half years. This study contributes to the existing knowledge in two aspects: first, it extends the use of an adaptive learning cognitive map and shows positive effects in supporting the implementation of the whole curriculum; second, it presents a long-term and large-scale example of successfully integrating ICTs into the key-competence-based curriculum reform in the Chinese context.

Video-based teacher online learning enables teachers to engage in reflective practice by watching others' classroom videos, providing peer feedback (PF) and reviewing others' work. However, the quality and reliability of PF often suffer due to variations in teaching proficiency among providers, which limits its usefulness for reviewers. To improve the quality of PF, hybrid intelligence is a promising approach that enhances human evaluation with AI-based techniques. Thus, we developed a hybrid intelligence feedback (HIF) system, where PF is categorized and summarized by large language models (LLM), and accompanied with AI multimodal data analysis, all in accordance with a knowledge structure. To investigate the effectiveness of the HIF, we conducted a study involving 58 pre-service mathematics teachers. After their initial feedback provision on a classroom video, they were divided into two groups. One group received HIF, while the other received traditional PF. Both groups revised their initial feedback based on the same video with PF or HIF report, and were assigned two tasks, namely in-depth reflection and extensive reflection. We analysed the reflective writings generated in in-depth reflection using the Structure of Observed Learning Outcomes taxonomy, and examined the diversity of teachers' attentions in extensive reflection using information entropy. Compared to traditional PF, our findings indicated that HIF (a) facilitated more comments added in feedback revision, (b) promoted multi-structural and extended abstract level thinking in in-depth reflection, (c) encouraged more diverse attentions in extensive reflection. These results demonstrate the effectiveness of HIF in enhancing PF to promote reviewers' reflection. This efficacy can be attributed to the utilization of LLM to identify common elements within PF, guided by the human knowledge-based framework, as well as the integration of data-driven evidence to complement PF.Practitioner notes What is already known about this topic? Video-based teacher online learning allows teachers to reflect on their own or others' videos flexibly while providing and reviewing peer feedback using reflection tools. While the benefits of reflecting on one's own videos with peer feedback are widely recognized, there is limited empirical evidence supporting the advantages of reflecting on others' videos with peer feedback. The effectiveness of this process may be affected by the quality and reliability of the peer feedback provided. Using natural language processing techniques to enhance peer feedback can be feasible and effective. However, it is primarily used to address textual-level issues and is less effective in promoting professionalism. Multimodal data analysis has shown effectiveness in enhancing teaching behaviours and facilitating reflection. However, despite the widespread use of AI-based analysis on classroom videos, they often lack educational interpretations. Hybrid intelligence is a novel concept in learning science research, aiming to leverage both human and machine intelligence to enhance the overall effectiveness. What this paper adds? This study applied the concept of hybrid intelligence to video-based teacher learning by proposing a hybrid intelligence feedback (HIF) system, aiming to promote teachers' reflection on others' videos during the peer feedback review process. This study proposed the design of HIF module, where human peer feedback was enhanced by large language models, and machine data analysis was complemented with educational interpretations, all structured according to an expert knowledge structure. The results showed that the HIF was effective to stimulate teachers' higher-level thinking in in-depth reflection and enhanced the diversity of their attentions in extensive reflection. However, it may still be challenging for novices to comprehend and integrate newly noticed pedagogical strategies in the HIF with their internal knowledge structures during reflection. Implications for practice and/or policy With the rapid advancement of generative artificial intelligence, the utilization of large language models becomes more flexible and effective, enabling multitasking enhancement for peer feedback in collaboration with human's professional knowledge. Multimodal data analysis effectively collaborates with human observations by managing low-level observation aspects, allowing humans to concentrate on higher-level thinking guided by the educational interpretations. The effectiveness of the HIF system is influenced by teachers' pedagogical knowledge, prior feedback provision experience and data literacy. In the future research, these diversities need to be taken into account in the design of video-based PD incorporating HIF to assess its long-term efficacy.

Teacher networks and communities have played an important role in teacher professional development. In such contexts, teachers often receive extensive feedback from peers as part of social learning. However, many teachers have difficulty identifying essential information from a large amount of peer feedback, which may impede self-reflection and peer learning. This study proposes the use of computer-assisted visual analytics of peer feedback to address this challenge. A visualisation-based analytical dashboard was designed and applied to help teachers analyse and reflect on peer feedback in a lesson study community for Technological Pedagogical Content Knowledge (TPACK) development. Primary school teachers participated in the lesson study, in which they collaborated to discuss lesson plans, observe recorded lessons, and give peer ratings and comments using an online platform. By comparing the performance between those using the analytical dashboard and others not using it, the results show that the approach has promising effects on improving teachers' TPACK as reflected in the lesson plans and their perceived confidence in TPACK. The implications of the findings are also discussed.

在智能时代,爆炸式增长的信息与人脑有限的认知能力之间的鸿沟日益扩大,人们需要借助外部设备来增强自身的认知能力,以更好地适应复杂的现代社会,认知外包成为时代发展的必然.文章阐述了认知外包的机理及其关键过程,并根据认知外包过程中外部设备在计算、感知、认知和社会化上不同层次提供的支持将认知外包分为了四种类型.而有效认知外包的核心在于内外部认知网络的平衡与有效连接,以及意义的理解和生成,这需要人们具备一些关键能力.在教育领域,如果学生自身头脑缺乏足够完备的知识和相应的关键能力,认知外包可能会导致其陷入典型教育陷阱:思维懒惰与幼稚化、认知地位边缘化、发展主体性丧失、认知肤浅与碎片化、认知偏见与极端化.为了帮助学生跨越这些陷阱,需要推进核心素养导向的深度教学变革.

This study aims to evaluate Chinese short-answer questions according to the Structure of Observed Learning Outcomes (SOLO) taxonomy, which divides students' responses into five levels, from prestructural to extended abstract. To automate the evaluation process, supervised classification was used. For this purpose, three deep-learning models were adopted: LSTM, C-LSTM, Nezha-BiLSTM. Results showed that the accuracy of LSTM was the lowest at 0.83, the accuracy of C-LSTM was 0.89, and the accuracy of Nezha-BiLSTM was the highest at 0.90. And the accuracy of each classification model is high in the prestructural and multistructural responses. The method proposed could help teachers perform scoring more efficiently.

Peer feedback is crucial to peer learning among teachers in a lesson study. However, teachers often have difficulties providing high-quality feedback on peer performance, which may influence their engagement in peer feedback practice. It remains unknown how teachers can be guided to provide meaningful instead of general or superficial feedback to other teachers' performance. To address the gap, this study proposed a framework guiding teachers to give meaningful feedback by focusing on core components of classroom teaching. A quasi-experimental study was conducted with 60 primary school teachers in an online lesson study. The results reveal promising effects of the approach on improving peer feedback quality and teacher engagement in peer feedback practice in an online lesson study. The teachers using the proposed approach outperformed those not using the approach in providing meaningful comments on questioning and assessment practice in classroom teaching, and in providing constructive comments on peers' weaknesses by giving explanations with references to theories. Although this study was conducted with teachers of Chinese as a local language, the findings have implications for professional development among teachers of Chinese as a foreign or second language.

In comparison to children and young students, adult learners usually exhibit more complex learning behaviors and psychological needs during the learning process. Designing social robots for adult learners has, thus, been a challenging task and a far less explored area, and it requires the great efforts from both technical and theoretical perspectives. We, therefore, first propose a novel framework that exploits the latest artificial intelligence (AI) technologies and the established psychological theory for social robot design. Under the proposed framework, we implement a novel social robot and deploy it in the challenging learning context, which demands the robot provide natural interactions and autonomous learning supports to adult learners. The evaluation results show that the robot significantly improves the learners' intrinsic motivation, and the adult learners have also shown great interests in learning and communicating with the robot. This article sheds light on how to design interactive and autonomous social robots for adult learners and contributes a concrete solution that employs the established psychological theories as the design guidelines and the AI models as the enabling technologies.

提供具有公共服务性质的在线教育来完善教育服务体系,成为后疫情时代教育服务供给改革的方向之一.由于在线教育公共服务所涉及的参与主体在属性上具有多元性,治理理论成为学界在分析相关问题时会援用的思考角度.本文以治理理论为基础,以北京市"开放辅导"为例,通过分析"开放辅导"的价值目标、多元化主体角色,以及在在线教育服务提供与制度优化等关键环节中蕴含的运行机制,探讨应该如何组织个性化的在线教育公共服务供给.本文认为,建构基于数据的教育服务供给的关键是强化在线教育在回应个性化学习需求方面的优势,要借助数据实现角色职能与互动机制的创新,构建更大范围内的协同机制、培育更深厚的协同意识,推动不同层面的教育服务供给主体与接受者进入良性的互动循环,以释放多元主体的优势,促进在线教育服务供给健康发展.

Most online learning researchers use resource recommendation and retrieve based on learning performance and learning style to provide accurate learning resources, but it is a closed and passive adaptive way. Learners always do not know the recommendation rationale and just receive the result-oriented recommended resources without having a chance to make a procedural reflection and self-directed adjustment. This study proposed an adaptive learning cognitive map model and developed an online learning system based on this model. The proposed system could represent learners' knowledge structure and cognitive state that allow learning contents, learning activities, learning paths and learning partners to be adjusted continuously, thus providing learners with appropriate learning resources and initiative reflection opportunities during the whole learning procedure. This study conducted a contrast experiment to evaluate the effectiveness of the proposed system. Students in the experimental group received learning interventions according to their adaptive learning cognitive maps. The result demonstrated that the proposed system with the learning cognitive map improved students' learning achievement, learning satisfied, user acceptance of information technology and learning interaction while did not reduce or increase cognitive load.

This paper introduces a system that supports student-centered online one-to-one tutoring and evaluates the practical value of the system by running an experiment with 64 experienced mathematics teachers and 810 students in Grade 7. The experiment lasted for 50 days. A comprehensive evaluation was performed using students' academic performance before and after usage of the system and the system log files. By classifying the students into active and inactive usage groups, it was determined that active students significantly outperformed inactive students on posttests, but with a small effect size. The results also suggested that high prior knowledge students tended to benefit more from using the system than low prior knowledge students. An explanation for this result was that students with a high level of prior knowledge were more likely to have good-quality interactions with their teachers. Therefore, although some advantages of this type of student-centered online one-to-one tutoring are observed, in this system, both the students and the teachers need to be further facilitated to produce more effective tutoring interactions.

A linked data approach provides new opportunities for annotating, interlinking, sharing and enriching massive open online educational resources. However, it can be difficult for non-expert users to build and utilize the educational linked data in educational settings. Thus, flexible and user-friendly ways to represent, interlink, visualize and utilize the educational linked data become increasingly important. This paper proposes a linked data-based knowledge navigation system (LDKNS) for improving the teaching effectiveness. In the system, learning resource ontology was implemented to represent learning contents, and linked data visualization technologies were used in a formal curriculum structure. It presents a use case of building the educational linked data and creating interactive knowledge visualizations to support teachers and students in art design education. A contrast experiment was conducted to evaluate the effectiveness of the proposed system. The experiment involved 115 second year undergraduate students divided into experimental and control groups. It was found that there was a significant increase in the motivation of students who used the LDKNS. It was also found that students’ achievement in the experimental group performs better. Furthermore, the results of a survey on cognitive load revealed that using the system can decrease their cognitive load. Thus, we believe that art design education supplemented with the LDKNS yields a significant learning advantage for students by improving learning performance.

Knowledge tracing (KT) refers to the issue of predicting learners' knowledge states based on their learning history and is the core technology for computer-assisted adaptive learning. The latest KT research has improved prediction performance by exploring the relationship between concepts and questions. However, these models have their parameters fixed after training. Without fully or partially re-training, they can only predict all learners' performance at all time steps using the same set of parameters, making them challenging to adapt to learners' personal development and individual differences. To address these problems, we take inspiration from synaptic plasticity, the primary neural mechanism conferring biological brains with lifelong learning capabilities, and propose the plastic gating network (PGN) to adapt to variation in learners' cognition over time and across individuals. At the core of the proposed model are plastic weights, which can constantly evolve after training according to the ongoing inputs and outputs. We adopt plastic weights in calculating gates and cell input in recurrent units, thus allowing the model to develop time- and individual-specific parameters that adapt to learners' personal development and individual differences after training. Experiments on real-world datasets show that the proposed PGN outperforms baseline methods in predicting the future performance of learners. Crucially, the PGN exhibits effective adaptation to learners' personal development and individual differences.

多模态大模型逐渐成为人工智能领域研究的热点,目前已在通用领域有显著进展,但在教育领域仍处于起步阶段.文章提出可以构建教育领域通用大模型,并使其通过下游任务适配形成三类多模态教育大模型,从而形成三种典型教育应用,即教学资源自动生成、人机协同过程支持与教师教学智能辅助.在此基础上,文章以"多模态汉字学习系统"为例,利用多模态大模型实现跨模态释义生成,展示了多模态大模型在辅助语言学习方面的应用潜力.最后,文章针对教育领域通用大模型研究、多模态教育大模型的创新应用及其带来的潜在风险与可能触发的教育变革,提出针对性的建议与展望.