IEEE Transactions on Learning Technologies最新文献

{"title":"MTGKT: Multiscale Temporal Graph Knowledge Tracing","authors":"Dong Liu;Chenghao Luo;Daolong Li;Yiliu Tu;Enhong Chen","doi":"10.1109/TLT.2026.3707679","DOIUrl":"https://doi.org/10.1109/TLT.2026.3707679","url":null,"abstract":"Knowledge tracing (KT) dynamically monitors students’ evolving knowledge mastery states through historical interaction sequences and predicts future performance. Existing KT methods often assume homogeneous temporal effects, and therefore, fail to fully capture the heterogeneous impact of multiscale intervals on knowledge dependence evolution. To tackle this challenge, we propose the multiscale temporal graph KT (MTGKT) model. First, we design a multiscale temporal encoder that decomposes intervals into micro-, meso-, and macroscales and fuses them through soft assignment. Second, we introduce an adaptive forgetting mechanism based on the Ebbinghaus forgetting curve, with both knowledge-specific and student-specific parameters to model individualized memory decay. Third, we devise an enhanced dynamic graph neural network with structured neighbor sampling, skill cooccurrence features, and message passing, together with gated recurrent unit and multihead attention for temporal dynamics. In experiments, we conducted comparative evaluations between MTGKT and nine representative KT models on four public datasets, and the results show competitive performance under the reported experimental settings.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"663-678"},"PeriodicalIF":7.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148626787","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Robots in Inclusive Education of Students With Autism Spectrum Disorder","authors":"Kaloyan Damyanov","doi":"10.1109/TLT.2026.3696075","DOIUrl":"https://doi.org/10.1109/TLT.2026.3696075","url":null,"abstract":"This study examines how humanoid and social robots affect social and communication skills in inclusive education for children with autism spectrum disorder (ASD). A mixed-methods approach is used to collect and analyze data using quantitative and qualitative methods. The study includes 30 children with ASD who use NAO, Pepper, and KASPAR robots for 12 weeks. The main task is to diagnose children’s social sensitivity and communication skills before and after using robots in education to evaluate this approach. Children’s social adaptation and robot-mediated interaction are associated with significant improvements, particularly in nonverbal communication and social interaction. Robots can make learning more inclusive and supportive, according to the study. The study examines robotic therapy development theory, including its pros and cons. The analysis focuses on successful ASD teaching methods in Norway, Japan, Bulgaria, and the USA, where inclusive education and advanced technologies are used to support students with special needs. Based on the data, practical recommendations are developed for teachers on how to integrate robots into teaching, for policymakers on equal access to technology, and for developers of robotic systems that focus on the needs of children with ASD.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"535-547"},"PeriodicalIF":4.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148236175","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A TextGrad-Inspired Optimization Pipeline for AI-Assisted Generation of Lecture Notes and Problem Sets","authors":"Chi Chiu So;Cheuk Ho Lee;Dan Yang;Anthony Wai Keung Loh;Chufang Wu","doi":"10.1109/TLT.2026.3731978","DOIUrl":"https://doi.org/10.1109/TLT.2026.3731978","url":null,"abstract":"Large language models (LLMs) hold substantial promise for AI-assisted education, yet challenges such as hallucinations and pedagogical misalignment hinder their direct application. This article introduces a pipeline adapting textual gradient-based optimization for the automated generation and iterative refinement of lecture notes and problem sets. Addressing LLM limitations through a seven-step framework inspired by mathematical gradient descent, the proposed pipeline operates as an instructional copilot. Grounded in rubrics aligned with Bloom’s Revised Taxonomy, the framework utilizes course outlines and textbooks to ensure controllable content refinement, structural coherence, and cognitive engagement. Evaluated across a Science, Technology, Engineering, and Mathematics (STEM) course (“Numerical Methods”) and a non-STEM course (“Introduction to Macroeconomics”), the proposed pipeline significantly outperformed ablation variants (including zero-shot prompting and unguided iterative refinement) in educator-rated quality. Quantitative analyses demonstrate rapid self-correction, systematic generation of interactive elements, and robust component indispensability. A complementary student pilot study provided preliminary evidence of positive learning gains and user receptivity. This framework reduces educator workload and mitigates generative flaws, paving the way for scalable teaching tools while emphasizing the necessity of human oversight in AI-driven pedagogy.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"907-924"},"PeriodicalIF":7.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11683899","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149130918","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Personalized Exercise Recommendation via Federated Learning With Hierarchical Attention","authors":"Weiwei Zhang;Siyu Chen;Zhaoyang Sun;Baicheng Ju","doi":"10.1109/TLT.2026.3730333","DOIUrl":"https://doi.org/10.1109/TLT.2026.3730333","url":null,"abstract":"Personalized exercise recommendation aims to provide students with suitable learning materials based on their individual knowledge states and learning needs. However, existing recommendation systems still face two challenges: they often overlook the alignment between exercise difficulty and student proficiency, and they usually rely on a shared model that cannot adequately capture the heterogeneity among different schools. In this article, we propose FedPER, a federated personalized exercise recommendation framework that integrates federated learning with a hierarchical attention mechanism to support personalized recommendations across heterogeneous educational environments. In FedPER, each school is treated as an independent client with a two-stage local model consisting of knowledge tracing and exercise recommendation. For knowledge tracing, a graph convolutional network is employed to model the relationships among knowledge concepts based on a Phi-coefficient correlation matrix, while a gated recurrent unit captures the dynamic changes in students' knowledge states over time. Based on the estimated knowledge states, the exercise recommendation module identifies students' weak concepts and selects exercises by jointly considering knowledge relevance and difficulty adaptation. On the server side, a hierarchical attention mechanism (FedHAM) is introduced to perform adaptive model aggregation by considering layerwise differences between global and local parameters, thereby improving the model's ability to handle nonindependent and nonidentically distributed educational data. Experiments conducted on the ASSISTments 2009 dataset show that FedPER achieves competitive performance in terms of difficulty-adaptation accuracy, which measures how closely the difficulty of the recommended exercises matches each student's ability, and novelty compared with existing approaches under both federated and nonfederated settings. The results demonstrate that FedPER can provide effective personalized exercise recommendations while adapting to the heterogeneous data distributions and learning characteristics of different school-level clients.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"877-889"},"PeriodicalIF":7.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149133514","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Design and Evaluation of an AI-Based Conversational Agent for Adaptive Feedback in Educational Robotics","authors":"Martina Morano;Dirk Ifenthaler;Lorenzo Cesaretti;Francesco Di Nardo;Laura Screpanti;David Scaradozzi","doi":"10.1109/TLT.2026.3676621","DOIUrl":"https://doi.org/10.1109/TLT.2026.3676621","url":null,"abstract":"Artificial intelligence is increasingly adopted in education to deliver structured, real-time feedback, especially in complex, collaborative learning scenarios. Large language models (LLMs), in particular, have enabled the integration of chatbots into a wide range of educational settings. This study presents the design and implementation of an AI-empowered chatbot, based on the Anthropic Claude 3.5 Sonnet, within educational robotics activities, aiming to explore its potential as a valid support system for real-time, contextualized feedback delivery. The system was deployed in a field study involving 80 students (grades 5–7), organized into 26 collaborative groups. Three hypotheses were evaluated: whether chatbot interaction correlates with students’ perceived usefulness (PU), supports more efficient programming behavior, and is related to task success. While the number of attempts did not significantly decrease, chatbot interaction showed mixed associations with success, highlighting that usage frequency alone does not predict effectiveness. Students also consistently rated the system as highly useful (PU = 4.04 <inline-formula><tex-math>$pm$</tex-math></inline-formula> 0.46). In addition, the chatbot infrastructure enables continuous tracking of student activity, opening the way for advanced student modeling and the development of digital twins within educational systems. This work contributes to the field of intelligent tutoring by demonstrating that LLM-based agents, when pedagogically aligned, can improve both perceived feedback quality and learner engagement in collaborative, task-based learning environments. Future research should refine feedback strategies and evaluate system performance across more homogeneous learner groups to improve modeling accuracy and scalability.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"302-315"},"PeriodicalIF":4.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147737067","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Interdisciplinary Integration: The Connotation and Methods of Cultivating Artificial Intelligence Talents in K-12 Education","authors":"Junjie Ni;Yadan Hua;Weiwei Wang;Junfeng Yang;Tingsong Li","doi":"10.1109/TLT.2026.3672830","DOIUrl":"https://doi.org/10.1109/TLT.2026.3672830","url":null,"abstract":"The development and use of artificial intelligence (AI) create a huge demand on talents with comprehensive AI skills that should be cultivated in K-12 education. However, the AI program in K-12 education remains a critical bottleneck in the AI-talent pipeline. This study posits that the K-12 stage should establish the “Five Core Competencies”—encompassing interdisciplinary integration, top-tier innovation, future-oriented development, foundational literacy and values, and AI fundamentals and core capabilities—as its unified objective. Grounded in Vygotsky’s theory of the Zone of Proximal Development, the study proposes a dynamic weighting of these competencies across different school phases. Building upon this, an “internal–external dual-cycle” cultivation architecture is constructed: the internal cycle operates through the iterative closed loop of “Curriculum system–AI tools–Teacher training” to drive student growth, while the external cycle functions via the synergistic closed loop of “Environment incentives–Management guarantee–Resource integration” to form the cultivation platform, thereby enabling dynamic talent development. This article designs a large-language-model (LLM) prompting framework oriented toward the five competencies to automatically generate interdisciplinary cultivation plans. The educational effectiveness of these AI-generated plans is then rigorously evaluated through a comparative experiment. The evaluation metrics are theoretically grounded in the works of Piaget, Bruner, and Gardner, focusing on the three dimensions of cognitive appropriateness, knowledge structuring, and intellectual diversity. The experimental results affirm that interdisciplinary integration constitutes a pivotal pathway for cultivating AI talent. Furthermore, the dual-cycle architecture, when coupled with the competency-guided LLM generation approach, demonstrably enhances the systematicity and adaptability of the cultivation plans.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"330-340"},"PeriodicalIF":4.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147796011","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The Integration of Interactive Psychoeducation and Agent Embodiment in Virtual Reality Training: An Evaluation on Breathing Techniques","authors":"Christian Corrò;Luca Chittaro","doi":"10.1109/TLT.2026.3683471","DOIUrl":"https://doi.org/10.1109/TLT.2026.3683471","url":null,"abstract":"Immersive virtual reality (VR) is increasingly used in education for its ability to create engaging learning experiences. However, VR-based systems for skill acquisition, especially in the health domain, often lack adequate educational foundations, potentially limiting their effectiveness. To address this limitation, this article investigates an interactive psychoeducation that incorporates users’ real-time physiological signals. Moreover, given the growing use of virtual agents in education and their potential to enhance psychoeducation delivery through natural human-like interactions, this article also studies the role of virtual agent embodiment as a second variable. We conducted a 2 × 2 between-subjects study (<italic>n</i> = 56) using a VR-based breathing training system, varying the presence of psychoeducation and virtual agent embodiment to assess their individual and combined impacts on users’ self-efficacy, motivation, attention, social presence, and performance of the taught skill. Results revealed that psychoeducation significantly enhanced self-efficacy and motivation, particularly by increasing participants’ interest in, and perceived value of, the experience, while agent embodiment improved social presence, relatedness, and skill performance. These findings show that interactive psychoeducation and embodied virtual agents can enhance complementary aspects of VR-based training effectiveness.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"404-417"},"PeriodicalIF":4.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11480742","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147937440","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ConfuciusLM: Exploring Confucian Personalized Instruction for Primary Learners With Large Language Models","authors":"Xin Xu;Xinya Lu;Yusi Yan;Mao Tian;Jianan Wang","doi":"10.1109/TLT.2026.3688630","DOIUrl":"https://doi.org/10.1109/TLT.2026.3688630","url":null,"abstract":"Large language models (LLMs) have emerged as a transformative force in educational informatization, offering unprecedented capabilities to simulate teacher–student interactions and generate adaptive instructional resources. However, general-purpose LLMs often provide direct outcome-oriented responses that lack pedagogical sensitivity, while existing educational models frequently prioritize rigid logical inquiry (e.g., Socratic questioning) over the affective and cognitive needs of novice learners. Such limitations can induce learning anxiety and impede the development of student confidence, particularly at the primary education level. To bridge this gap, we propose ConfuciusLM, a specialized educational LLM fine-tuned on a novel instructional dialogue dataset grounded in Confucian pedagogical philosophy. ConfuciusLM embodies a learner-centered paradigm by implementing a heuristic reasoning framework that integrates cognitive guidance with affective support and motivational scaffolding. Through both automated metrics and human-centered pedagogical evaluations, we demonstrate that ConfuciusLM exhibits superior performance compared to open-source LLMs of similar parameter scales in fostering positive learning dispositions and maintaining instructional coherence. This study advances the field of LLM-empowered education by proposing a robust framework that integrates Eastern pedagogical principles with artificial-intelligence-driven instructional design, thereby informing the development of more inclusive, empathetic, and human-centered intelligent tutoring systems. Importantly, the proposed approach offers concrete pedagogical and design implications for primary education, supporting the creation of learning environments that are psychologically safe, conducive to the development of self-efficacy, and capable of sustaining learner engagement under appropriate teacher guidance.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"433-446"},"PeriodicalIF":4.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147937631","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Scenario-Driven AI+VR Language Practice Platform: A Technical Feasibility and Formative Usability Study","authors":"Zeynep Cipiloglu Yildiz;Furkan Baytak;Furkan Ozkaya","doi":"10.1109/TLT.2026.3679468","DOIUrl":"https://doi.org/10.1109/TLT.2026.3679468","url":null,"abstract":"In this article, we present a scenario-driven artificial intelligence (AI) + virtual reality (VR) language practice platform designed to support immersive role-play and open-ended spoken interaction for L2 speaking practice. The system combines interactive VR scenarios, large-language-model-driven real-time conversations, a block-based scenario authoring tool for customizing practice content, reflective analytics derived from conversation logs, and a privacy-aware social layer for peer support. The platform integrates the OpenAI GPT-4 Realtime application programming interface (API) within the Unity 3D engine to enable low-latency conversational interaction in immersive environments. We report a technical feasibility assessment and a formative usability evaluation. In a study with 30 learners, the platform achieved a mean System Usability Scale score of 73.25, a p95 frame time of 10.69 ms, and an average AI response latency of 0.77 s. These results provide initial evidence that the proposed design can deliver responsive AI-mediated dialogue and a usable end-to-end experience for scenario-based practice. We conclude with design tradeoffs, key limitations of the current short-term evaluation, and directions for learning-oriented and longitudinal studies.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"286-301"},"PeriodicalIF":4.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147696214","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An Explicit-Thinking LLM-Based Cognitive Apprenticeship Approach to Promote Students’ Metacognition, Higher Order Thinking Skills, and Learning Performance","authors":"Yihua Zhong;Jili Chen;Changqin Huang;Xizhe Wang;Yixuan Chen;Tao He","doi":"10.1109/TLT.2026.3696968","DOIUrl":"https://doi.org/10.1109/TLT.2026.3696968","url":null,"abstract":"With the rapid development of generative artificial intelligence (AI), large language models (LLMs) have been increasingly applied in education. Although LLMs demonstrate considerable potential in supporting student learning, conventional LLMs tend to provide direct answers without explicitly articulating their reasoning processes. This hinders students’ ability to comprehend the underlying logic of problem solving, thereby limiting opportunities for metacognitive engagement and the development of higher order thinking skills. To address this issue, this study proposes a cognitive apprenticeship approach based on an explicit-thinking LLM that provides structured thinking support during LLM-based learning activities. By employing the modeling, coaching, and reflection mechanisms of cognitive apprenticeship, the approach guides students to internalize the LLM’s reasoning processes, thereby fostering their higher order thinking skills. A quasi-experimental design was adopted, involving 68 college students divided into experimental and control groups. Participants completed a writing-based learning task using either the explicit-thinking LLM or a conventional LLM, and the data were analyzed using a mixed-methods approach that included eye tracking. The results showed that, compared to the control group, the experimental group exhibited enhanced metacognitive regulation, demonstrated superior performance in both problem solving and critical thinking, and achieved better learning outcomes when they devoted greater attention to the reasoning process generated by the explicit-thinking LLM. This study suggests that the explicit-thinking LLM enhances students’ learning performance and higher order thinking by making reasoning processes more transparent. It further emphasizes the essential role of cultivating students’ critical and creative engagement with AI to effectively support the development of higher order thinking skills.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"19 ","pages":"548-565"},"PeriodicalIF":4.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148236176","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}