arXiv subjects

Unggi Lee

Publications and source records attributed to Unggi Lee.

At least 19 recordsLinked to original sources

arXiv2026

EduClaw-Bench: A Long-Horizon Benchmark for Pedagogical LLM Agents with Simulated Learners

Large language models (LLMs) power educational applications from tutoring to essay scoring, but each is a point solution to a single task, and only recently have these point solutions been integrated into agents operating over a learning management system (LMS). Yet tutoring is long-horizon, since a learner improves over days and weeks rather than in a single turn, and no benchmark evaluates an agent tutor across a sustained relationship. We introduce EduClaw-Bench, a benchmark that places an agent tutor in a continuous 30-day relationship with a simulated learner grounded in knowledge tracing (KT), whose knowledge-concept mastery, from a KT model trained on real-student data, drives its answers and is probed for learning gain across 55 scenarios. Each agent is scored on three primary axes (learning gain, responsiveness, and helpfulness) and two curriculum-design axes (Gagn\'e and Rosenshine), with helpfulness and the curriculum axes judged by a cross-family panel of three LLM judges. Evaluating 10 agent adapters over three base-model tiers yields two findings that single-tier, single-session evaluation cannot reach. First, tutoring quality belongs to the base model and the agent harness together rather than either alone. Second, almost no combination sustains good tutoring over the full horizon. A calibration check ($\text{ECE}=0.049$) and a live-classroom field study confirm that the simulated learner and its measurements track reality. Our work is a step toward trustworthy AI tutors for future education.

cs.CY↗

arXiv2026

Not the Dimension, the Norm: What Matters in Gradient-Free Weight Perturbation of Language Models

Adapting a language model to a task no longer requires training all of its weights, and a line of parameter-efficient methods has driven the trainable count from billions down to a handful of scalars. Gradient-free adaptation, which samples random weight perturbations and keeps the ones that score well, has not followed that trajectory and still perturbs every entry of the weight tensor. It is unknown whether that full-weight search is necessary, and more fundamentally which property of a perturbation makes it work at all, because existing methods vary the search space, the perturbation scale, and the aggregation together. We resolve this by intervening on one factor at a time inside a fixed pipeline, holding candidate scoring and voting constant while we vary the search dimension, the subspace that carries the perturbation, and its norm. Perturbing a frozen frame of 12 to 16 scalars stays 1.8 accuracy points behind full-weight search on average across 49 model-benchmark cells, trailing it in 36 of them. Neither the dimension nor the choice of basis explains that performance. A random frame whose Grassmann overlap with the SVD frame is at chance level performs identically once a single scale factor is matched, and at large scales the SVD directions collapse first. What survives is the perturbation norm, whose usable range closes within a factor of five across seven models and stays flat inside. The perturbation norm is therefore the one factor with a failure mode, and its safe region transfers across scale and family. The design question narrows from which subspace to perturb to how hard to shake.

cs.CL↗

arXiv2026

Reinforcement Learning for Special Education: Aligning LLM Tutors to Diverse Learners through Disability-Adaptive Training

Large language models are increasingly deployed as intelligent tutors, yet research on aligning them for special education remains absent. Recent work has applied reinforcement learning to LLM tutors, but these methods target a generic learner in a single domain (mathematics) and do not address the cognitive and communicative diversity of learners with disabilities. We introduce \emph{Special-R1}, a framework that extends pedagogical RL to special education through two components: (1) a two-dimensional adaptive system prompt that couples a difficulty-based support level with a disability-specific teaching style across five disability profiles; and (2) a persona-aware Thinking Reward whose judge rubric is conditioned on the learner's disability profile. On a persona-augmented test set of 690 multi-turn dialogues, our full model raises persona-aware Fit from 6.75 (generic baseline) to 8.40 (+1.65) and SPED-rubric Helpfulness from 0.720 to 0.768, leading on the four-component Total (2.911, +0.064 over the runner-up) while remaining within 0.01 of the strongest variant on the out-of-domain OpenLearnLM benchmark (8.53). Ablations show that the Thinking Reward becomes effective only in combination with adaptive prompting, and that residual weakness on specific learning disability in mathematics motivates targeted multimodal extensions.

cs.CY↗

arXiv2026

TeachObs: A Human-Validated Benchmark for Multimodal Teaching Observation and Model Evaluation

Classroom videos contain observable teaching practices, but their pedagogical and visual signals are rarely organized in forms suitable for model evaluation. We present \textit{TeachObs}, a human-validated benchmark for multimodal teaching observation in classroom videos. \textit{TeachObs} includes 30 public lesson videos from eight countries divided into 5,158 fixed 15-second scenes. Seven researchers annotated each scene with 39 binary observation codes, covering 20 visual codes, such as gesture, board work, pointing, and visual materials, and 19 nonvisual codes, such as instruction, monitoring, questioning, feedback, and reflection. Gold segment labels are constructed using reliability- and prevalence-aware rules based on Krippendorff's alpha. In addition to segment-level labels, three expert raters produced lesson-level ratings and qualitative evaluations of instructional design, instructional delivery, learner response, learning materials, and lesson closure across the 30 lessons, with rater coverage detailed in the body. Using these two human reference layers, we evaluate five vision-capable frontier LLMs across three tracks - text-only segment coding, text + frame segment coding, and lesson-level coverage scored under an LLM-as-judge protocol - and find that no single model consistently outperforms others across all three tracks, that adding a mid-frame inflates both true and false attributions per scene, and that model evaluations over-rate procedurally clear lessons relative to expert raters. \textit{TeachObs} therefore supports both fine-grained annotation benchmarking and whole-lesson evaluation, showing where AI systems can assist classroom video analysis and where expert judgment remains necessary across varied subjects, classroom formats, and annotation difficulty levels.

cs.CL↗

arXiv2026

The Tutoring Effectiveness Index: Predicting LLM Math Tutor Quality from Four Conversation Signals

Aligning large language models (LLMs) as math tutors typically demands costly reinforcement-learning (RL) training and external LLM judges. We ask whether a frozen model's internal reasoning signals can replace both. We propose the Tutoring Effectiveness Index (TEI), a training-free, judge-free four-signal index that combines a Schoenfeld-Verify keyword ratio, a math-step density, an ends-question rate, and a deep-reasoning gate from the Deep-Thinking Ratio (DTR) probe. Selecting from $N$ candidates with TEI (the TEI@$N$ rule) raises the improvement rate on pre-incorrect scenarios from $59.0\%$ to $81.9\%$ at $N{=}8$ on a frozen DeepSeek-R1-8B base, with no training and no external judge. We also measure the alignment tax of pedagogical GRPO. Thinking length drops from $1{,}764$ to $119$ words per turn ($-93\%$), Content-Knowledge and Pedagogical-Knowledge accuracy fall by $-71\%$ and $-80\%$ relative, and the student's $\Delta$ Solve Rate crosses from $+0.180$ to $-0.012$. To anchor the behavioural reading, we reproduce an 82-code educational codebook on $119{,}009$ tutor sentences with a one-shot structural classifier. Together, these results offer a cost-effective recipe for building math-tutoring LLMs without RL training or external judges.

cs.CY↗

arXiv2026

BuddyBench: A Privacy-Constrained Multi-Task Benchmark for Pediatric Social-Communication Personalization

BuddyBench introduces a privacy-constrained multi-task benchmark for pediatric social-communication personalization. Unlike existing neurodevelopmental repositories that primarily emphasize imaging, genetics, or cross-sectional clinical phenotyping, BuddyBench links drill-level learning trajectories, standardized clinical assessments, BuddyPlan self-report, and randomized-treatment endpoints within a unified benchmark schema. BuddyBench combines two cohorts: ND-03 is an observational cohort with dense drill coverage for Tasks1-2 (n = 189), and ND-02 is a randomized controlled trial cohort for Tasks3-4 (n = 86 ITT). Together, they support knowledge tracing, next-drill recommendation, clinical prediction, and causal inference, linking behavioral personalization to clinical evaluation. We additionally introduce BuddyBench-Sim, a synthetic companion dataset for reproducible evaluation. Baselines show signal across tasks while keeping pediatric clinical records protected.

cs.AI↗

arXiv2026

Are Video Models Zero-Shot Learners and Reasoners in Education? EduVideoBench, A Knowledge-Skills-Attitude Benchmark for Educational Video Generation

Video generation models (VGMs) are rapidly entering classrooms, yet existing benchmarks evaluate only perceptual quality, intrinsic faithfulness, generic safety, or video as a reasoning medium, and none assesses whether the outputs are educationally valid. In this work, we present EduVideoBench, the first balanced benchmark in the education domain, grounded in the Knowledge-Skills-Attitude (KSA) framework so that pedagogical adequacy and educational safety are evaluated jointly rather than as ad-hoc quality dimensions. Across five frontier VGMs, our results show substantial room for improvement across knowledge, skills, and attitude before they are classroom-ready. We complement this with a qualitative analysis of expert comments, finding that educational validity is multi-component, where a single misaligned element such as pacing, legibility, or notation can invalidate an otherwise correct video. We hope EduVideoBench will guide the development of VGMs that are pedagogically grounded and safe for the classroom.

cs.CL↗

arXiv2026

LLMs Are Already Good Tutors: Training-Free Prompt Optimization for Pedagogical Math Tutoring

Aligning LLMs for math tutoring typically requires RL-based training with multi-GPU infrastructure. We investigate whether training-free prompt optimization-evolving only the system prompt via API calls-can serve as a practical alternative. We adapt 7 published methods and propose 5 education-specialized methods, evaluating these 12 methods under 5 conditions on 2 OOD benchmark suites. All 12 best-per-method configurations surpass the strongest RL-trained baseline (R_total = 0.633), and our ParetoGrad achieves the best Pareto balance across post-test solve rate, leak control, and helpfulness, rather than dominating any single component. Behavioral analysis with an 82-code educational codebook reveals that training-free methods rely on teaching-knowledge patterns at 2-3x the rate of RL-trained models, with a compensating ~10 percentage-point reduction in intent-level scaffolding. We also find a task-dependent reasoning mode effect consistent across training-free and RL-based paradigms. Our approach enables efficient development of pedagogically aligned LLM tutors with prompts alone and minimal compute.

cs.CL↗

arXiv2026

Llama-Polya: Instruction Tuning for Large Language Model based on Polya's Problem-solving

This paper introduces Llama-Polya, an instruction-tuned large language model that integrates Polya's four-step problem-solving framework into its dialogue structure to support mathematical reasoning. Mathematical problem-solving is central to students' success in mathematics education, yet many learners struggle to plan, justify, and verify their solutions. Although large language models (LLMs) show promise as intelligent tutors, they often lack structured pedagogical alignment grounded in established learning theories. To address this gap, we operationalize Polya's problem-solving framework within an instruction-tuned LLM to promote metacognitive engagement and examine the effects of pedagogy-aligned fine-tuning compared to domain-only and general-purpose instruction tuning. Built on the Llama-3.1-8B architecture, Llama-Polya was fine-tuned on synthetic math problem-solving data derived from GSM8K, structured according to Polya's four stages. We developed and evaluated multiple variants-general-purpose instruct, math-domain metamath, pedagogy-aligned polya-v2, and sequential metamath+polya-v2-using both quantitative accuracy metrics and qualitative pedagogical assessments. Results indicate that models tuned with Polya's framework and domain-specific data produced more balanced reasoning-stage distributions and fewer premature answers. Expert evaluators also observed improved pedagogical coherence and metacognitive prompting, although limitations in personalization and mathematical rigor remained. These findings suggest that pedagogy-grounded instruction tuning can enhance educational alignment and reasoning transparency in LLM-based tutoring systems.

cs.CY↗

arXiv2026

ISD-Agent-Bench: A Comprehensive Benchmark for Evaluating LLM-based Instructional Design Agents

Large Language Model (LLM) agents have shown promising potential in automating Instructional Systems Design (ISD), a systematic approach to developing educational programs. However, evaluating these agents remains challenging due to the lack of standardized benchmarks and the risk of LLM-as-judge bias. We present ISD-Agent-Bench, a comprehensive benchmark comprising 25,795 scenarios generated via a Context Matrix framework that combines 51 contextual variables across 5 categories with 33 ISD sub-steps derived from the ADDIE model. To ensure evaluation reliability, we employ a multi-judge protocol using diverse LLMs from different providers, achieving high inter-judge reliability. We compare existing ISD agents with novel agents grounded in classical ISD theories such as ADDIE, Dick \& Carey, and Rapid Prototyping ISD. Experiments on 1,017 test scenarios demonstrate that integrating classical ISD frameworks with modern ReAct-style reasoning achieves the highest performance, outperforming both pure theory-based agents and technique-only approaches. Further analysis reveals that theoretical quality strongly correlates with benchmark performance, with theory-based agents showing significant advantages in problem-centered design and objective-assessment alignment. Our work provides a foundation for systematic LLM-based ISD research.

cs.SE↗

arXiv2026

Exploring the Role of Automated Feedback in Programming Education: A Systematic Literature Review

Automated feedback systems have become increasingly integral to programming education, where learners engage in iterative cycles of code construction, testing, and refinement. Despite its wider integration in practices and technical advancements into AI, research in this area remains fragmented, lacking synthesis across technological and instructional dimensions. This systematic literature review synthesizes 61 empirical studies published by September 2024, offering a conceptually grounded analysis of automated feedback systems across five dimensions: system architecture, pedagogical function, interaction mechanism, contextual deployment, and evaluation approach. Findings reveal that most systems are fully automated, embedded within online platforms, and primarily focused on error detection and code correctness. While recent developments incorporate adaptive features and large language models to enable more personalized and interactive feedback, few systems offer support for higher-order learning processes, interactive components, or learner agency. Moreover, evaluation practices tend to emphasize short-term performance gains, with limited attention to long-term outcomes or instructional integration. These findings call for a reimagining of automated feedback not as a technical add-on for error correction, but as a pedagogical scaffold that supports deeper, adaptive, and interactive learning.

cs.CY↗

arXiv2026

Rewarding How Models Think Pedagogically: Integrating Pedagogical Reasoning and Thinking Rewards for LLMs in Education

Large language models (LLMs) are increasingly deployed as intelligent tutoring systems, yet research on optimizing LLMs specifically for educational contexts remains limited. Recent works have proposed reinforcement learning approaches for training LLM tutors, but these methods focus solely on optimizing visible responses while neglecting the model's internal thinking process. We introduce PedagogicalRL-Thinking, a framework that extends pedagogical alignment to reasoning LLMs in education through two novel approaches: (1) Pedagogical Reasoning Prompting, which guides internal reasoning using domain-specific educational theory rather than generic instructions; and (2) Thinking Reward, which explicitly evaluates and reinforces the pedagogical quality of the model's reasoning traces. Our experiments reveal that domain-specific, theory-grounded prompting outperforms generic prompting, and that Thinking Reward is most effective when combined with pedagogical prompting. Furthermore, models trained only on mathematics tutoring dialogues show improved performance on educational benchmarks not seen during training, while preserving the base model's factual knowledge. Our quantitative and qualitative analyses reveal that pedagogical thinking reward produces systematic reasoning trace changes, with increased pedagogical reasoning and more structured instructional decision-making in the tutor's thinking process.

cs.CL↗

arXiv2026

Pedagogical Alignment for Vision-Language-Action Models: A Comprehensive Framework for Data, Architecture, and Evaluation in Education

Science demonstrations are important for effective STEM education, yet teachers face challenges in conducting them safely and consistently across multiple occasions, where robotics can be helpful. However, current Vision-Language-Action (VLA) models require substantial computational resources and sacrifice language generation capabilities to maximize efficiency, making them unsuitable for resource-constrained educational settings that require interpretable, explanation-generating systems. We present \textit{Pedagogical VLA Framework}, a framework that applies pedagogical alignment to lightweight VLA models through four components: text healing to restore language generation capabilities, large language model (LLM) distillation to transfer pedagogical knowledge, safety training for educational environments, and pedagogical evaluation adjusted to science education contexts. We evaluate Pedagogical VLA Framework across five science demonstrations spanning physics, chemistry, biology, and earth science, using an evaluation framework developed in collaboration with science education experts. Our evaluation assesses both task performance (success rate, protocol compliance, efficiency, safety) and pedagogical quality through teacher surveys and LLM-as-Judge assessment. We additionally provide qualitative analysis of generated texts. Experimental results demonstrate that Pedagogical VLA Framework achieves comparable task performance to baseline models while producing contextually appropriate educational explanations.

cs.CL↗

arXiv2026

OpenLearnLM Benchmark: A Unified Framework for Evaluating Knowledge, Skill, and Attitude in Educational Large Language Models

Large Language Models are increasingly deployed as educational tools, yet existing benchmarks focus on narrow skills and lack grounding in learning sciences. We introduce OpenLearnLM Benchmark, a theory-grounded framework evaluating LLMs across three dimensions derived from educational assessment theory: Knowledge (curriculum-aligned content and pedagogical understanding), Skills (scenario-based competencies organized through a four-level center-role-scenario-subscenario hierarchy), and Attitude (alignment consistency and deception resistance). Our benchmark comprises 124K+ items spanning multiple subjects, educational roles, and difficulty levels based on Bloom's taxonomy. The Knowledge domain prioritizes authentic assessment items from established benchmarks, while the Attitude domain adapts Anthropic's Alignment Faking methodology to detect behavioral inconsistency under varying monitoring conditions. Evaluation of seven frontier models reveals distinct capability profiles: Claude-Opus-4.5 excels in practical skills despite lower content knowledge, while Grok-4.1-fast leads in knowledge but shows alignment concerns. Notably, no single model dominates all dimensions, validating the necessity of multi-axis evaluation. OpenLearnLM provides an open, comprehensive framework for advancing LLM readiness in authentic educational contexts.

cs.CL↗

arXiv2026

A Training-Free Large Reasoning Model-based Knowledge Tracing Framework for Unified Prediction and Prescription

Knowledge Tracing (KT) aims to estimate a learner's evolving mastery based on interaction histories. Recent studies have explored Large Language Models (LLMs) for KT via autoregressive nature, but such approaches typically require fine-tuning and exhibit unstable or near-random performance. Moreover, prior KT systems primarily focus on prediction and rely on multi-stage pipelines for feedback and recommendation, resulting in increased system complexity and resources. To address this gap, we propose Thinking-KT, a training-free KT framework that incorporates Test-Time Scaling (TTS), enabling even small LLMs to achieve competitive KT performance. Moreover, in this framework, a small LLM can jointly perform KT prediction, personalized feedback generation, and learning recommendation in a unified output without degrading prediction accuracy. Beyond performance, we present the systematic analysis of reasoning traces in KT. Our results demonstrate that TTS is a critical yet underexplored factor in LLM-based KT, and that small LLMs can serve as unified ITS engines.

cs.CL↗

arXiv2025

Pedagogy-R1: Pedagogically-Aligned Reasoning Model with Balanced Educational Benchmark

Recent advances in large reasoning models (LRMs) show strong performance in structured domains such as mathematics and programming; however, they often lack pedagogical coherence and realistic teaching behaviors. To bridge this gap, we introduce Pedagogy-R1, a framework that adapts LRMs for classroom use through three innovations: (1) a distillation-based pipeline that filters and refines model outputs for instruction-tuning, (2) the Well-balanced Educational Benchmark (WBEB), which evaluates performance across subject knowledge, pedagogical knowledge, tracing, essay scoring, and teacher decision-making, and (3) a Chain-of-Pedagogy (CoP) prompting strategy for generating and eliciting teacher-style reasoning. Our mixed-method evaluation combines quantitative metrics with qualitative analysis, providing the first systematic assessment of LRMs' pedagogical strengths and limitations.

cs.AI↗

arXiv2025

How Can Video Generative AI Transform K-12 Education? Examining Teachers' Perspectives through TPACK and TAM

The rapid advancement of generative AI technology, particularly video generative AI (Video GenAI), has opened new possibilities for K-12 education by enabling the creation of dynamic, customized, and high-quality visual content. Despite its potential, there is limited research on how this emerging technology can be effectively integrated into educational practices. This study explores the perspectives of leading K-12 teachers on the educational applications of Video GenAI, using the TPACK (Technological Pedagogical Content Knowledge) and TAM (Technology Acceptance Model) frameworks as analytical lenses. Through interviews and hands-on experimentation with video generation tools, the research identifies opportunities for enhancing teaching strategies, fostering student engagement, and supporting authentic task design. It also highlights challenges such as technical limitations, ethical considerations, and the need for institutional support. The findings provide actionable insights into how Video GenAI can transform teaching and learning, offering practical implications for policy, teacher training, and the future development of educational technology.

cs.CY↗

arXiv2025

Echo-Teddy: Preliminary Design and Development of Large Language Model-based Social Robot for Autistic Students

Autistic students often face challenges in social interaction, which can hinder their educational and personal development. This study introduces Echo-Teddy, a Large Language Model (LLM)-based social robot designed to support autistic students in developing social and communication skills. Unlike previous chatbot-based solutions, Echo-Teddy leverages advanced LLM capabilities to provide more natural and adaptive interactions. The research addresses two key questions: (1) What are the design principles and initial prototype characteristics of an effective LLM-based social robot for autistic students? (2) What improvements can be made based on developer reflection-on-action and expert interviews? The study employed a mixed-methods approach, combining prototype development with qualitative analysis of developer reflections and expert interviews. Key design principles identified include customizability, ethical considerations, and age-appropriate interactions. The initial prototype, built on a Raspberry Pi platform, features custom speech components and basic motor functions. Evaluation of the prototype revealed potential improvements in areas such as user interface, educational value, and practical implementation in educational settings. This research contributes to the growing field of AI-assisted special education by demonstrating the potential of LLM-based social robots in supporting autistic students. The findings provide valuable insights for future developments in accessible and effective social support tools for special education.

cs.HC↗