arXiv subjects

Minsun Kim

Publications and source records attributed to Minsun Kim.

At least 19 recordsLinked to original sources

arXiv2026

TutorTrace: A Dataset and Taxonomy for Classifying Learner Behavioral States during AI-Assisted Programming Education

AI programming tutors provide scalable support, yet lack the behavioral context human tutors rely on to adapt support to learners' needs. We present TutorTrace, a dataset and behavioral abstraction pipeline that makes learners' behavioral context visible and computable in real time from low-level IDE telemetry. Across four deployments in two introductory Python courses (N=480), TutorTrace captures approximately 180K telemetry events, 13,633 behavioral segments, and 27 continuously computed metrics. From this foundation, we derive a taxonomy of learner activity before the first AI query, between consecutive queries, and across the full session, enabling systems to respond not just to what learners say, but to what they have done leading up to the help-seeking moment. In a preliminary classroom evaluation, behavior-aware prompts were associated with a decrease in intervals between queries with no independent work from 50.0% to 20.7%. As an additional demonstration of downstream utility, we evaluate TutorTrace on two held-out prediction tasks: whether a learner will query within the next 60 seconds (AUROC=.726) and whether an upcoming query reflects guided or dependent help-seeking (AUROC=.717). Together, these findings show how behavioral context can enable adaptive AI tutoring at scale.

cs.AI↗

arXiv2026

Exploring the Design Space of LLM-Based Programming Support in CS Education: A Scoping Review through the Lens of Assistance Governance

As large language models (LLMs) become integrated into programming education, learner-facing systems increasingly differ in how that assistance is bounded, enacted, and controlled. These governance decisions are often described implicitly, making it difficult to compare systems in educationally meaningful ways. To address this gap, we conduct a scoping review and qualitative synthesis of 90 peer-reviewed LLM-based programming support systems in CS education. We analyze assistance governance through three dimensions, which we refer to collectively as PEA: Policy, capturing what forms of help are allowed or restricted; Enforcement, capturing how those boundaries are operationalized through interaction and system behavior; and Authority, capturing who can configure, adapt, or override them during use. Our findings show that systems often share similar pedagogical goals, but implement those goals through varied enforcement mechanisms. At the same time, authority remains highly centralized in system logic, with fewer systems giving learners or instructors runtime control. This work contributes PEA as a three-dimensional analytic lens, a governance codebook empirically refined within these dimensions, and a map of underexplored configurations in the current design space of LLM-based programming support. By making these explicit and comparable, PEA offers a vocabulary for analyzing existing systems and designing future tools that are pedagogically bounded, configurable, and accountable.

cs.HC↗

arXiv2026

Student-ChatGPT Interaction Visible: Designing a Teacher Dashboard for EFL Writing Education

We present a Prompt Analytics Dashboard (PAD) for teachers that can traces student-LLM interactions from EFL writing classes. PAD can show student prompt-response exchanges with LLM chatbot and English essay writing revision histories to support data-informed instruction and visibility in classes. Through two iterative co-design sessions with six EFL instructors, we distilled a compact trace taxonomy (misuse signals, goal-alignment cues, revision effort) and instantiated three interface views (overview, week/outcome filter, drill-down with evidence snippets). This pipeline summarizes potential misuse and alignment at class/cohort levels and attaches micro-explanations to reduce over-surveillance. Instructors reported reduced scanning burden and clearer timing for interventions.

cs.HC↗

arXiv2026

ComVi: Context-Aware Optimized Comment Display in Video Playback

On general video-sharing platforms like YouTube, comments are displayed independently of video playback. As viewers often read comments while watching a video, they may encounter ones referring to moments unrelated to the current scene, which can reveal spoilers and disrupt immersion. To address this problem, we present ComVi, a novel system that displays comments at contextually relevant moments, enabling viewers to see time-synchronized comments and video content together. We first map all comments to relevant video timestamps by computing audio-visual correlation, then construct the comment sequence through an optimization that considers temporal relevance, popularity (number of likes), and display duration for comfortable reading. In a user study, ComVi provided a significantly more engaging experience than conventional video interfaces (i.e., YouTube and Danmaku), with 71.9% of participants selecting ComVi as their most preferred interface.

cs.MM↗

arXiv2024

LLM-Driven Learning Analytics Dashboard for Teachers in EFL Writing Education

This paper presents the development of a dashboard designed specifically for teachers in English as a Foreign Language (EFL) writing education. Leveraging LLMs, the dashboard facilitates the analysis of student interactions with an essay writing system, which integrates ChatGPT for real-time feedback. The dashboard aids teachers in monitoring student behavior, identifying noneducational interaction with ChatGPT, and aligning instructional strategies with learning objectives. By combining insights from NLP and Human-Computer Interaction (HCI), this study demonstrates how a human-centered approach can enhance the effectiveness of teacher dashboards, particularly in ChatGPT-integrated learning.

cs.HC↗

arXiv2024

Designing Prompt Analytics Dashboards to Analyze Student-ChatGPT Interactions in EFL Writing

While ChatGPT has significantly impacted education by offering personalized resources for students, its integration into educational settings poses unprecedented risks, such as inaccuracies and biases in AI-generated content, plagiarism and over-reliance on AI, and privacy and security issues. To help teachers address such risks, we conducted a two-phase iterative design process that comprises surveys, interviews, and prototype demonstration involving six EFL (English as a Foreign Language) teachers, who integrated ChatGPT into semester-long English essay writing classes. Based on the needs identified during the initial survey and interviews, we developed a prototype of Prompt Analytics Dashboard (PAD) that integrates the essay editing history and chat logs between students and ChatGPT. Teacher's feedback on the prototype informs additional features and unmet needs for designing future PAD, which helps them (1) analyze contextual analysis of student behaviors, (2) design an overall learning loop, and (3) develop their teaching skills.

cs.HC↗

arXiv2024

LLM-as-a-tutor in EFL Writing Education: Focusing on Evaluation of Student-LLM Interaction

In the context of English as a Foreign Language (EFL) writing education, LLM-as-a-tutor can assist students by providing real-time feedback on their essays. However, challenges arise in assessing LLM-as-a-tutor due to differing standards between educational and general use cases. To bridge this gap, we integrate pedagogical principles to assess student-LLM interaction. First, we explore how LLMs can function as English tutors, providing effective essay feedback tailored to students. Second, we propose three metrics to evaluate LLM-as-a-tutor specifically designed for EFL writing education, emphasizing pedagogical aspects. In this process, EFL experts evaluate the feedback from LLM-as-a-tutor regarding quality and characteristics. On the other hand, EFL learners assess their learning outcomes from interaction with LLM-as-a-tutor. This approach lays the groundwork for developing LLMs-as-a-tutor tailored to the needs of EFL learners, advancing the effectiveness of writing education in this context.

cs.CL↗

arXiv2024

ChEDDAR: Student-ChatGPT Dialogue in EFL Writing Education

The integration of generative AI in education is expanding, yet empirical analyses of large-scale, real-world interactions between students and AI systems still remain limited. In this study, we present ChEDDAR, ChatGPT & EFL Learner's Dialogue Dataset As Revising an essay, which is collected from a semester-long longitudinal experiment involving 212 college students enrolled in English as Foreign Langauge (EFL) writing courses. The students were asked to revise their essays through dialogues with ChatGPT. ChEDDAR includes a conversation log, utterance-level essay edit history, self-rated satisfaction, and students' intent, in addition to session-level pre-and-post surveys documenting their objectives and overall experiences. We analyze students' usage patterns and perceptions regarding generative AI with respect to their intent and satisfaction. As a foundational step, we establish baseline results for two pivotal tasks in task-oriented dialogue systems within educational contexts: intent detection and satisfaction estimation. We finally suggest further research to refine the integration of generative AI into education settings, outlining potential scenarios utilizing ChEDDAR. ChEDDAR is publicly available at https://github.com/zeunie/ChEDDAR.

cs.CL↗

arXiv2024

RECIPE4U: Student-ChatGPT Interaction Dataset in EFL Writing Education

The integration of generative AI in education is expanding, yet empirical analyses of large-scale and real-world interactions between students and AI systems still remain limited. Addressing this gap, we present RECIPE4U (RECIPE for University), a dataset sourced from a semester-long experiment with 212 college students in English as Foreign Language (EFL) writing courses. During the study, students engaged in dialogues with ChatGPT to revise their essays. RECIPE4U includes comprehensive records of these interactions, including conversation logs, students' intent, students' self-rated satisfaction, and students' essay edit histories. In particular, we annotate the students' utterances in RECIPE4U with 13 intention labels based on our coding schemes. We establish baseline results for two subtasks in task-oriented dialogue systems within educational contexts: intent detection and satisfaction estimation. As a foundational step, we explore student-ChatGPT interaction patterns through RECIPE4U and analyze them by focusing on students' dialogue, essay data statistics, and students' essay edits. We further illustrate potential applications of RECIPE4U dataset for enhancing the incorporation of LLMs in educational frameworks. RECIPE4U is publicly available at https://zeunie.github.io/RECIPE4U/.

cs.CL↗

arXiv2023

On the origin of optical and IR emission lines in star forming galaxies

Combining the {\sc Cloudy} photoionization code with updated stellar population synthesis results, we simultaneously model the MIR $\neiii/\neii$ vs. $\oiv/\neiii$, the MIR-FIR $\neiii/\neii$ vs. $\oiv/\oiii$ and the classical BPT diagnostic diagrams. We focus on the properties of optically classified \hii\,galaxies that lie in the normal star forming zone in the MIR diagnostic diagram. We find that a small fraction of our models lie in this zone, but most of them correspond to the lowest explored metallicity, \zstar\,=\,0.0002, at age $\sim1$ Gyr. This value of \zstar\,is, by far, lower than the values derived for these galaxies from optical emission lines, suggesting that the far-UV emission produced by post-AGB stars (a.k.a. HOLMES, hot low-mass evolved stars) is NOT the source of ionization. Instead, shock models can easily reproduce this part of the MIR diagram. We suggest that it is likely that some of these galaxies have been misclassified and that in them, shocks, produced by a weak AGN-outflow, could be an important source of ionizaton. Using a subset of our models, we derive a new demarcation line for the maximal contribution of retired galaxies in the BPT diagram. This demarcation line allows for a larger contamination from the neighbouring AGN-dominated region. Considering the importance of disentangling the different ionising mechanisms in weak or deeply obscured systems, new observational efforts to classify galaxies both in the optical and IR are required to better constrain this kind of models and understand their evolutionary paths.

astro-ph.GA↗

arXiv2023

RECIPE: How to Integrate ChatGPT into EFL Writing Education

The integration of generative AI in the field of education is actively being explored. In particular, ChatGPT has garnered significant interest, offering an opportunity to examine its effectiveness in English as a foreign language (EFL) education. To address this need, we present a novel learning platform called RECIPE (Revising an Essay with ChatGPT on an Interactive Platform for EFL learners). Our platform features two types of prompts that facilitate conversations between ChatGPT and students: (1) a hidden prompt for ChatGPT to take an EFL teacher role and (2) an open prompt for students to initiate a dialogue with a self-written summary of what they have learned. We deployed this platform for 213 undergraduate and graduate students enrolled in EFL writing courses and seven instructors. For this study, we collect students' interaction data from RECIPE, including students' perceptions and usage of the platform, and user scenarios are examined with the data. We also conduct a focus group interview with six students and an individual interview with one EFL instructor to explore design opportunities for leveraging generative AI models in the field of EFL education.

cs.HC↗

arXiv2021

A hyperparameter-tuning approach to automated inverse planning

Radiotherapy inverse planning often requires planners to modify parameters in the treatment planning system's objective function to produce clinically acceptable plans. Due to the manual steps in this process, plan quality can vary depending on the planning time available and the planner's skills. This study investigates two hyperparameter-tuning methods for automated inverse planning. Because this framework does not train a model on previously-optimized plans, it can be readily adapted to practice pattern changes, and plan quality is not limited by that of a training cohort. We selected 10 patients who received lung SBRT using manually-generated clinical plans. We used random sampling (RS) and Bayesian optimization (BO) to tune parameters using linear-quadratic utility functions based on 11 clinical goals. Normalizing all plans to have PTV D95 equal to 48 Gy, we compared plan quality for the automatically-generated and manually-generated plans. We also investigated the impact of iteration count on the automatically-generated plans, comparing planning time and plan utility for RS and BO plans with and without stopping criteria. Without stopping criteria, the median planning time was 1.9 and 2.3 hours for RS and BO plans. The OAR doses in the RS and BO plans had a median percent difference (MPD) of 48.7% and 60.4% below clinical dose limits and an MPD of 2.8% and 3.3% below clinical plan doses. With stopping criteria, the utility decreased by an MPD of 5.3% and 3.9% for RS and BO plans, but the median planning time was reduced to 0.5 and 0.7 hours, and the OAR doses still had an MPD of 42.9% and 49.7% below clinical dose limits and an MPD of 0.3% and 1.8% below clinical plan doses. This study demonstrates that hyperparameter-tuning approaches to automated inverse planning can reduce active planning time with plan quality that is similar to or better than manually-generated plans.

physics.med-ph↗

arXiv2021

A nonconvex optimization approach to IMRT planning with dose-volume constraints

Fluence map optimization for intensity-modulated radiation therapy planning can be formulated as a large-scale inverse problem with competing objectives and constraints associated with the tumors and organs-at-risk. Unfortunately, clinically relevant dose-volume constraints are nonconvex, so standard algorithms for convex problems cannot be directly applied. While prior work focused on convex approximations for these constraints, we propose a novel relaxation approach to handle nonconvex dose-volume constraints. We develop efficient, provably convergent algorithms based on partial minimization, and show how to adapt them to handle maximum-dose constraints and infeasible problems. We demonstrate our approach using the CORT dataset, and show that it is easily adaptable to radiation treatment planning with dose-volume constraints for multiple tumors and organs-at-risk.

math.OC↗

arXiv2019

A Proof of Principle: Multi-Modality Radiotherapy Optimization

Radiotherapy is used to treat cancer patients by damaging DNA of tumor cells using ionizing radiation. Photons are the most widely used radiation type for therapy, having been put into use soon after the first discovery of X-rays in 1895. However, there are emerging interests and developments of other radiation modalities such as protons and carbon ions, owing to their unique biological and physical characteristics that distinguish these modalities from photons. Current attempts to determine an optimal radiation modality or an optimal combination of multiple modalities are empirical and in the early stage of development. In this paper, we propose a mathematical framework to optimize full radiation dose distributions and fractionation schedules of multiple radiation modalities, aiming to maximize the damage to the tumor while limiting the damage to the normal tissue to the corresponding tolerance level. This formulation gives rise to a non-convex, mixed integer program and we propose a bilevel optimization algorithm, to efficiently solve it. The upper level problem is to optimize the fractionation schedule using the dose distribution optimized in the lower level. We demonstrate the feasibility of our novel framework and algorithms in a simple 2-dimensional phantom with two different radiation modalities, where clinical intuition can be easily drawn. The results of our numerical simulations agree with the clinical intuition, validating our approach and showing the promise of the framework for further clinical investigation.

math.OC↗

arXiv2017

A Markov decision process approach to optimizing cancer therapy using multiple modalities

There are several different modalities, e.g., surgery, chemotherapy, and radiotherapy, that are currently used to treat cancer. It is common practice to use a combination of these modalities to maximize clinical outcomes, which are often measured by a balance between maximizing tumor damage and minimizing normal tissue side effects due to treatment. However, multi-modality treatment policies are mostly empirical in current practice, and are therefore subject to individual clinicians' experiences and intuition. We present a novel formulation of optimal multi-modality cancer management using a finite-horizon Markov decision process approach. Specifically, at each decision epoch, the clinician chooses an optimal treatment modality based on the patient's observed state, which we define as a combination of tumor progression and normal tissue side effect. Treatment modalities are categorized as (1) Type 1, which has a high risk and high reward, but is restricted in the frequency of administration during a treatment course, (2) Type 2, which has a lower risk and lower reward than Type 1, but may be repeated without restriction, and (3) Type 3, no treatment (surveillance), which has the possibility of reducing normal tissue side effect at the risk of worsening tumor progression. Numerical simulations using various intuitive, concave reward functions show the structural insights of optimal policies and demonstrate the potential applications of using a rigorous approach to optimizing multi-modality cancer management.

math.OC↗

arXiv2016

On the Evolution of High-Redshift Active Galactic Nuclei

We build a simple physical model to study the high-redshift active galactic Nucleus (AGN) evolution within the co-evolution framework of central black holes (BHs) and their host galaxies. The correlation between the circular velocity of a dark halo $V_c$ and the velocity dispersion of a galaxy $σ$ is used to link the dark matter halo mass and BH mass. The dark matter halo mass function is converted to the BH mass function for any given redshift. The high-redshift optical AGN luminosity functions (LFs) are constructed. At $z\sim 4$, the flattening feature is not shown at the faint end of the optical AGN LF. This is consistent with observational results. If the optical AGN LF at $z\sim 6$ can be reproduced in the case in which central BHs have the Eddington-limited accretion, it is possible for the AGN lifetime to have a small value of $2\times 10^5$ yrs. The X-ray AGN LFs and X-ray AGN number counts are also calculated at $2.0 3$, respectively, using the same parameters adopted in the calculation for the optical AGN LF at $z\sim 4$. It is estimated that about 30 AGNs per $\rm{deg}^2$ at $z>6$ can be detected with a flux limit of $3\times 10^{-17}~\rm{erg~cm^{-2}~s^{-1}}$ in the $0.5-2$ keV band. Additionally, the cosmic reionization is also investigated. The ultraviolet photons emitted from the high-redshift AGNs mainly contribute to the cosmic reionization, and the central BHs of the high-redshift AGNs have a mass range of $10^6-10^8M_\odot$. We also discuss some uncertainties in both the AGN LFs and AGN number counts originating from the $M_{\rm{BH}}-σ$ relation, Eddington ratio, AGN lifetime, and X-ray attenuation in our model.

astro-ph.GA↗

arXiv2012

A Hierachical Evolutionary Algorithm for Multiobjective Optimization in IMRT

Purpose: Current inverse planning methods for IMRT are limited because they are not designed to explore the trade-offs between the competing objectives between the tumor and normal tissues. Our goal was to develop an efficient multiobjective optimization algorithm that was flexible enough to handle any form of objective function and that resulted in a set of Pareto optimal plans. Methods: We developed a hierarchical evolutionary multiobjective algorithm designed to quickly generate a diverse Pareto optimal set of IMRT plans that meet all clinical constraints and reflect the trade-offs in the plans. The top level of the hierarchical algorithm is a multiobjective evolutionary algorithm (MOEA). The genes of the individuals generated in the MOEA are the parameters that define the penalty function minimized during an accelerated deterministic IMRT optimization that represents the bottom level of the hierarchy. The MOEA incorporates clinical criteria to restrict the search space through protocol objectives and then uses Pareto optimality among the fitness objectives to select individuals. Results: Acceleration techniques implemented on both levels of the hierarchical algorithm resulted in short, practical runtimes for optimizations. The MOEA improvements were evaluated for example prostate cases with one target and two OARs. The modified MOEA dominated 11.3% of plans using a standard genetic algorithm package. By implementing domination advantage and protocol objectives, small diverse populations of clinically acceptable plans that were only dominated 0.2% by the Pareto front could be generated in a fraction of an hour. Conclusions: Our MOEA produces a diverse Pareto optimal set of plans that meet all dosimetric protocol criteria in a feasible amount of time. It optimizes not only beamlet intensities but also objective function parameters on a patient-specific basis.

physics.med-ph↗

arXiv2010

A Wide Field Survey of Satellite Galaxies around the Spiral Galaxy M106

We present a wide field survey of satellite galaxies in M106 (NGC 4258) covering a $1.7\degr \times 2\degr$ field around M106 using Canada-France-Hawaii Telescope/MegaCam. We find 16 satellite galaxy candidates of M106. Eight of these galaxies are found to be dwarf galaxies that are much smaller and fainter than the remaining galaxies. Four of these galaxies are new findings. Surface brightness profiles of 15 out of 16 satellite galaxies can be represented well by an exponential disk profile with varying scale length. We derive the surface number density distribution of these satellite galaxies. The central number density profile (d $<100$ kpc) is well fitted by a power-law with a power index of $-2.1\pm0.5$, similar to the expected power index of isothermal distribution. The luminosity function of these satellites is represented well by the Schechter function with a faint end slope of $-1.19^{+0.03}_{-0.06}$. Integrated photometric properties (total luminosity, total colour, and disk scale length) and the spatial distribution of these satellite galaxies are found to be roughly similar to those of the Milky Way and M31.

astro-ph.CO↗