arXiv subjects

Yixin Cheng

Publications and source records attributed to Yixin Cheng.

12 recordsLinked to original sources

arXiv2026

Dataset of GenAI-Assisted Information Problem Solving in Education

Information Problem Solving (IPS) is a critical competency for academic and professional success in education, work, and life. The advent of Generative Artificial Intelligence (GenAI), particularly tools like ChatGPT, has introduced new possibilities for supporting students in complex IPS tasks. However, empirical insights into how students engage with GenAI during IPS and how these tools can be effectively leveraged for learning remain limited. Moreover, differences in background--shaped by cultural and socioeconomic factors--pose additional challenges to the equitable integration of GenAI in educational contexts. To address this gap, we present an open-source dataset collected from 279 students at a public Australian university. The dataset was generated through students' use of FLoRA, a GenAI-powered educational platform that is widely adopted in the field of learning analytics. Within FLoRA, students interacted with an embedded GenAI chatbot to gather information and synthesize it into data science project proposals. The dataset captures fine-grained, multi-dimensional records of GenAI-assisted IPS processes, including: (i) student-GenAI dialogue transcripts; (ii) writing process log traces; (iii) final project proposals with human-assigned assessment scores; (iv) two surveys assessing students demographic background and their prior knowledge and experience in data science and AI; and (v) surveys capturing students' perceptions of GenAI's effectiveness in supporting IPS and platform use experience. This dataset provides a valuable resource for advancing our understanding of GenAI's role in educational IPS and informing the design of adaptive, inclusive AI-powered learning tools.

cs.HC↗

arXiv2025

Dean of LLM Tutors: A Framework for Automated Quality Review of AI-generated Feedback

Large language model (LLM) tutors are increasingly used to generate educational feedback, but existing research has focused mainly on feedback generation rather than feedback evaluation. As a result, LLM-generated feedback may offer limited pedagogical value and carry risks of hallucination. The current study introduces DeanLLM, an automated review framework for comprehensively evaluating feedback generated by LLM tutors before it is shared with students. We developed a 16-dimension evaluation framework covering feedback content, educational effectiveness, and hallucination risks, and validated it using using human-expert annotations of LLM-generated tutor feedback on synthetic computer science assignment submissions derived from real coursework. We then examined whether LLMs could serve as automated LLM-generated tutor feedback reviewers, and used the best-performing reviewer to benchmark tutor feedback generated by 10 commercial LLMs. Psychometric analyses supported the reliability of the proposed framework and showed that human reviewers tended to evaluate feedback holistically, whereas the LLM reviewer separated rubric dimensions more mechanically. Standard zero-shot and few-shot prompting showed limited agreement with human experts for content-quality judgments. Supervised fine-tuning of GPT-4.1 with human-labelled examples containing scores only, without explanatory rationales, achieved the strongest alignment with expert judgments. Reasoning LLMs were particularly effective at hallucination detection and produced automated tutor feedback with stronger educational effectiveness and factuality than lightweight models. The findings indicate that DeanLLM offers a scalable way for automatically improving the reliability and safety of LLM tutor feedback, while also demonstrating that reviewer calibration and model choice remain critical for educational deployment.

cs.CY↗

arXiv2025

SCALEFeedback: A Large-Scale Dataset of Synthetic Computer Science Assignments for LLM-generated Educational Feedback Research

Using Large Language Models (LLMs) to give educational feedback to students for their assignments has attracted much attention in the AI in Education (AIED) field. Yet, there is currently no large-scale open-source dataset of student assignments that includes detailed assignment descriptions, rubrics, and student submissions across various courses. As a result, research on generalisable methodology for automatic generation of effective and responsible educational feedback remains limited. In this paper, we introduce a synthetic computer science university assignment dataset for LLM-based educational feedback research, called SCALEFeedback (Synthetic Computer science Assignments for LLM Educational Feedback Research). The dataset is generated via Sophisticated Assignment Mimicry (SAM) framework specifically designed to synthesise this dataset and that utilizes one-to-one LLM-based imitation from real assignment descriptions, rubrics, and student submissions. Our open-source dataset contains 10,000 synthetic student submissions spanning 155 assignments across 59 university-level computer science courses. Technical validation confirmed that the synthetic dataset closely resembles real data while successfully eliminating personally identifiable information present in the source material. The creation of this dataset is a valuable contribution to researchers who aim to develop LLM-based generalisable methods for offering high-quality, automated educational feedback in a scalable way.

cs.CY↗

arXiv2025

Revealing Weaknesses in Text Watermarking Through Self-Information Rewrite Attacks

Text watermarking aims to subtly embed statistical signals into text by controlling the Large Language Model (LLM)'s sampling process, enabling watermark detectors to verify that the output was generated by the specified model. The robustness of these watermarking algorithms has become a key factor in evaluating their effectiveness. Current text watermarking algorithms embed watermarks in high-entropy tokens to ensure text quality. In this paper, we reveal that this seemingly benign design can be exploited by attackers, posing a significant risk to the robustness of the watermark. We introduce a generic efficient paraphrasing attack, the Self-Information Rewrite Attack (SIRA), which leverages the vulnerability by calculating the self-information of each token to identify potential pattern tokens and perform targeted attack. Our work exposes a widely prevalent vulnerability in current watermarking algorithms. The experimental results show SIRA achieves nearly 100% attack success rates on seven recent watermarking methods with only 0.88 USD per million tokens cost. Our approach does not require any access to the watermark algorithms or the watermarked LLM and can seamlessly transfer to any LLM as the attack model, even mobile-level models. Our findings highlight the urgent need for more robust watermarking.

cs.LG↗

arXiv2024

Self-regulated Learning Processes in Secondary Education: A Network Analysis of Trace-based Measures

While the capacity to self-regulate has been found to be crucial for secondary school students, prior studies often rely on self-report surveys and think-aloud protocols that present notable limitations in capturing self-regulated learning (SRL) processes. This study advances the understanding of SRL in secondary education by using trace data to examine SRL processes during multi-source writing tasks, with higher education participants included for comparison. We collected fine-grained trace data from 66 secondary school students and 59 university students working on the same writing tasks within a shared SRL-oriented learning environment. The data were labelled using Bannert's validated SRL coding scheme to reflect specific SRL processes, and we examined the relationship between these processes, essay performance, and educational levels. Using epistemic network analysis (ENA) to model and visualise the interconnected SRL processes in Bannert's coding scheme, we found that: (a) secondary school students predominantly engaged in three SRL processes -- Orientation, Re-reading, and Elaboration/Organisation; (b) high-performing secondary students engaged more in Re-reading, while low-performing students showed more Orientation process; and (c) higher education students exhibited more diverse SRL processes such as Monitoring and Evaluation than their secondary education counterparts, who heavily relied on following task instructions and rubrics to guide their writing. These findings highlight the necessity of designing scaffolding tools and developing teacher training programs to enhance awareness and development of SRL skills for secondary school learners.

cs.HC↗

arXiv2024

Ink and Algorithm: Exploring Temporal Dynamics in Human-AI Collaborative Writing

The advent of Generative Artificial Intelligence (GAI) has revolutionized the field of writing, marking a shift towards human-AI collaborative writing in education. However, the dynamics of human-AI interaction in the collaborative writing process are not well understood, and thus it remains largely unknown how human learning can be effectively supported with such cutting-edge GAI technologies. In this study, we aim to bridge this gap by investigating how humans employ GAI in collaborative writing and examining the interplay between the patterns of GAI usage and human writing behaviors. Considering the potential varying degrees to which people rely on GAI usage, we proposed to use Dynamic Time Warping time-series clustering for the identification and analysis of common temporal patterns in AI usage during the human-AI collaborative writing processes. Additionally, we incorporated Epistemic Network Analysis to reveal the correlation between GAI usage and human writing behaviors that reflect cognitive processes (i.e., knowledge telling, knowledge transformation, and cognitive presence), aiming to offer insights for developing better approaches and tools to support human to learn effectively via such human-AI collaborative writing activities. Our findings reveal four major distinct temporal patterns in AI utilization and highlight significant correlations between these patterns and human writing behaviors. These findings have significant implications for effectively supporting human learning with GAI in educational writing tasks.

cs.HC↗

arXiv2024

A Study of Posterior Stability for Time-Series Latent Diffusion

Latent diffusion has demonstrated promising results in image generation and permits efficient sampling. However, this framework might suffer from the problem of posterior collapse when applied to time series. In this paper, we first show that posterior collapse will reduce latent diffusion to a variational autoencoder (VAE), making it less expressive. This highlights the importance of addressing this issue. We then introduce a principled method: dependency measure, that quantifies the sensitivity of a recurrent decoder to input variables. Using this tool, we confirm that posterior collapse significantly affects time-series latent diffusion on real datasets, and a phenomenon termed dependency illusion is also discovered in the case of shuffled time series. Finally, building on our theoretical and empirical studies, we introduce a new framework that extends latent diffusion and has a stable posterior. Extensive experiments on multiple real time-series datasets show that our new framework is free from posterior collapse and significantly outperforms previous baselines in time series synthesis.

cs.LG↗

arXiv2024

Leveraging the Context through Multi-Round Interactions for Jailbreaking Attacks

Large Language Models (LLMs) are susceptible to Jailbreaking attacks, which aim to extract harmful information by subtly modifying the attack query. As defense mechanisms evolve, directly obtaining harmful information becomes increasingly challenging for Jailbreaking attacks. In this work, inspired from Chomsky's transformational-generative grammar theory and human practices of indirect context to elicit harmful information, we focus on a new attack form, called Contextual Interaction Attack. We contend that the prior context\u2014the information preceding the attack query\u2014plays a pivotal role in enabling strong Jailbreaking attacks. Specifically, we propose a first multi-turn approach that leverages benign preliminary questions to interact with the LLM. Due to the autoregressive nature of LLMs, which use previous conversation rounds as context during generation, we guide the model's question-response pair to construct a context that is semantically aligned with the attack query to execute the attack. We conduct experiments on seven different LLMs and demonstrate the efficacy of this attack, which is black-box and can also transfer across LLMs. We believe this can lead to further developments and understanding of security in LLMs.

cs.LG↗

arXiv2024

Multilinear Operator Networks

Despite the remarkable capabilities of deep neural networks in image recognition, the dependence on activation functions remains a largely unexplored area and has yet to be eliminated. On the other hand, Polynomial Networks is a class of models that does not require activation functions, but have yet to perform on par with modern architectures. In this work, we aim close this gap and propose MONet, which relies solely on multilinear operators. The core layer of MONet, called Mu-Layer, captures multiplicative interactions of the elements of the input token. MONet captures high-degree interactions of the input elements and we demonstrate the efficacy of our approach on a series of image recognition and scientific computing benchmarks. The proposed model outperforms prior polynomial networks and performs on par with modern architectures. We believe that MONet can inspire further research on models that use entirely multilinear operations.

cs.CV↗

arXiv2024

Evidence-centered Assessment for Writing with Generative AI

We propose a learning analytics-based methodology for assessing the collaborative writing of humans and generative artificial intelligence. Framed by the evidence-centered design, we used elements of knowledge-telling, knowledge transformation, and cognitive presence to identify assessment claims; we used data collected from the CoAuthor writing tool as potential evidence for these claims; and we used epistemic network analysis to make inferences from the data about the claims. Our findings revealed significant differences in the writing processes of different groups of CoAuthor users, suggesting that our method is a plausible approach to assessing human-AI collaborative writing.

cs.HC↗

arXiv2022

Improving Students' Academic Performance with AI and Semantic Technologies

Artificial intelligence and semantic technologies are evolving and have been applied in various research areas, including the education domain. Higher Education institutions strive to improve students' academic performance. Early intervention to at-risk students and a reasonable curriculum is vital for students' success. Prior research opted for deploying traditional machine learning models to predict students' performance. In terms of curriculum semantic analysis, after conducting a comprehensive systematic review regarding the use of semantic technologies in the Computer Science curriculum, a major finding of the study is that technologies used to measure similarity have limitations in terms of accuracy and ambiguity in the representation of concepts, courses, etc. To fill these gaps, in this study, three implementations were developed, that is, to predict students' performance using marks from the previous semester, to model a course representation in a semantic way and compute the similarity, and to identify the prerequisite between two similar courses. Regarding performance prediction, we used the combination of Genetic Algorithm and Long-Short Term Memory (LSTM) on a dataset from a Brazilian university containing 248730 records. As for similarity measurement, we deployed BERT to encode the sentences and used cosine similarity to obtain the distance between courses. With respect to prerequisite identification, TextRazor was applied to extract concepts from course description, followed by employing SemRefD to measure the degree of prerequisite between two concepts. The outcomes of this study can be summarized as: (i) a breakthrough result improves Manrique's work by 2.5% in terms of accuracy in dropout prediction; (ii) uncover the similarity between courses based on course description; (iii) identify the prerequisite over three compulsory courses of School of Computing at ANU.

cs.CY↗

arXiv2022

The use of Semantic Technologies in Computer Science Curriculum: A Systematic Review

Semantic technologies are evolving and being applied in several research areas, including the education domain. This paper presents the outcomes of a systematic review carried out to provide an overview of the application of semantic technologies in the context of the Computer Science curriculum and discuss the limitations in this field whilst offering insights for future research. A total of 4,510 studies were reviewed, and 37 were analysed and reported. As a result, while semantic technologies have been increasingly used to develop Computer Science curricula, the alignment of ontologies and accurate curricula assessment appears to be the most significant limitations to the widespread adoption of such technologies.

cs.CY↗