arXiv subjects

Chunyang Chen

Publications and source records attributed to Chunyang Chen.

At least 37 records · Page 2Linked to original sources

arXiv2026

Exploring MLLMs Perception of Network Visualization Principles

In this paper, we test whether Multimodal Large Language Models (MLLMs) can match human-subject performance in tasks involving the perception of properties in network layouts. Specifically, we replicate a human-subject experiment about perceiving quality (namely stress) in network layouts using GPT-4o, Gemini-2.5 and Qwen2.5. Our experiments show that giving MLLMs the same study information as trained human participants yields performance comparable to that of human experts and exceeds that of untrained non-experts. Additionally, we show that prompt engineering that deviates from the human-subject experiment can lead to better-than-human performance in some settings. Interestingly, like human subjects, the MLLMs seem to rely on visual proxies rather than computing the actual value of stress, indicating some sense or facsimile of perception. Explanations from the models are similar to those used by the human participants (e.g., an even distribution of nodes and uniform edge lengths).

cs.HC↗

arXiv2026

RippleGUItester: Change-Aware Exploratory Testing

Software systems evolve continuously through frequent code changes, yet such changes often introduce unintended bugs despite extensive testing and code review. Existing testing approaches are largely constrained to predefined execution paths or rely on unguided exploration, leaving many change-induced issues undetected. To address this challenge, we present RippleGUItester, a change-driven testing system that treats a code change as the epicenter of a ripple effect and explores its broader, user-visible impacts via the GUI. Given a code change, RippleGUItester performs LLM-based change-impact analysis to generate and enrich realistic test scenarios, executes these scenarios on both pre-change and post-change versions of the system, and applies differential analysis to identify behavioral differences. Crucially, RippleGUItester employs multimodal bug detection, comparing visual GUI changes and interpreting them in the context of natural-language change intents to distinguish unintended bugs from intended behavioral updates. We evaluate our approach on hundreds of real-world code changes across four widely used software systems: Firefox, Zettlr, JabRef, and Godot. Our results show that the proposed approach uncovers bugs introduced by code changes that were missed by existing test suites, CI pipelines, and code review. In total, we identify 26 previously unknown bugs that still exist in the latest versions of the evaluated systems. After reporting, 16 bugs have been fixed, 2 have been confirmed, 6 are still under discussion, and 2 were marked as intended. We envision RippleGUItester being applied before or shortly after a code change is merged, enabling earlier detection of regressions.

cs.SE↗

arXiv2026

How Smart Is Your GUI Agent? A Framework for the Future of Software Interaction

GUI agents are rapidly becoming a new interaction to software, allowing people to navigate web, desktop and mobile rather than execute them click by click. Yet ``agent'' is described with radically different degrees of autonomy, obscuring capability, responsibility and risk. We call for conceptual clarity through GUI Agent Autonomy Levels (GAL), a six-level framework that makes autonomy explicit and helps benchmark progress toward trustworthy software interaction.

cs.SE↗

arXiv2026

SOGPTSpotter: Detecting ChatGPT-Generated Answers on Stack Overflow

Stack Overflow is a popular Q&A platform where users ask technical questions and receive answers from a community of experts. Recently, there has been a significant increase in the number of answers generated by ChatGPT, which can lead to incorrect and unreliable information being posted on the site. While Stack Overflow has banned such AI-generated content, detecting whether a post is ChatGPT-generated remains a challenging task. We introduce a novel approach, SOGPTSpotter, that employs Siamese Neural Networks, leveraging the BigBird model and the Triplet loss, to detect ChatGPT-generated answers on Stack Overflow. We use triplets of human answers, reference answers, and ChatGPT answers. Our empirical evaluation reveals that our approach outperforms well-established baselines like GPTZero, DetectGPT, GLTR, BERT, RoBERTa, and GPT-2 in identifying ChatGPT-synthesized Stack Overflow responses. We also conducted an ablation study to show the effectiveness of our model. Additional experiments were conducted to assess various factors, including the impact of text length, the model's robustness against adversarial attacks, and its generalization capabilities across different domains and large language models. We also conducted a real-world case study on Stack Overflow. Using our tool's recommendations, Stack Overflow moderators were able to identify and take down ChatGPT-suspected generated answers, demonstrating the practical applicability and effectiveness of our approach.

cs.SE↗

arXiv2025

A Semantic-based Optimization Approach for Repairing LLMs: Case Study on Code Generation

Language Models (LMs) are widely used in software engineering for code generation, but they may produce erroneous code. Rather than repairing outputs, a more thorough remedy is to address underlying model failures. LM repair offers a lightweight solution: it requires minimal data, lowers computational cost, and limits side effects. Unlike full retraining, LM repair focuses on applying tailored updates to targeted neurons, making it suitable for limited resources, high-performance demands, or strict safety requirements. In this paper, we propose Semantic Targeting for Analytical Repair (STAR), a novel semantic-based optimization method for repairing LLMs. STAR realizes the main operations of repairing LMs in an optimization process, including locating ``buggy neurons'', solving ``neuron patches'', and patching ``buggy neurons''. The neuron patches are computed with a solid semantic-based analytical formula, which directly bridges the changes to logits with the deltas of neurons, by steering latent representations. Compared to the prior work of LM repair (MINT) and standard optimization methods (SGD), STAR integrates their strengths while mitigating their limitations. By reformulating LM repair as an optimization process, STAR may solve multiple failures together, significantly improving the usefulness. Evaluated on coding tasks using popular code LMs, STAR demonstrates superior effectiveness compared with the state-of-the-art. Besides, STAR exhibits better efficiency. In terms of side effects, namely the balance between generalization and specificity, STAR outperforms prior work by a significant margin. Additionally, we conducted assessments on the overfitting risk of LM repair as well as the cumulative impact. Further, we analyzed the differences with pipeline-based methods and explained the reason why STAR is better and how it mitigated the common limitations of LM repair.

cs.SE↗

arXiv2025

A Longitudinal Measurement of Privacy Policy Evolution for Large Language Models

Large language model (LLM) services have been rapidly integrated into people's daily lives as chatbots and agentic systems. They are nourished by collecting rich streams of data, raising privacy concerns around excessive collection of sensitive personal information. Privacy policies are the fundamental mechanism for informing users about data practices in modern information privacy paradigm. Although traditional web and mobile policies are well studied, the privacy policies of LLM providers, their LLM-specific content, and their evolution over time remain largely underexplored. In this paper, we present the first longitudinal empirical study of privacy policies for mainstream LLM providers worldwide. We curate a chronological dataset of 74 historical privacy policies and 115 supplemental privacy documents from 11 LLM providers across 5 countries up to August 2025, and extract over 3,000 sentence-level edits between consecutive policy versions. We compare LLM privacy policies to those of other software formats, propose a taxonomy tailored to LLM privacy policies, annotate policy edits and align them with a timeline of key LLM ecosystem events. Results show they are substantially longer, demand college-level reading ability, and remain highly vague. Our taxonomy analysis reveals patterns in how providers disclose LLM-specific practices and highlights regional disparities in coverage. Policy edits are concentrated in first-party data collection and international/specific-audience sections, and that product releases and regulatory actions are the primary drivers, shedding light on the status quo and the evolution of LLM privacy policies.

cs.CR↗

arXiv2025

ReTrace: Interactive Visualizations for Reasoning Traces of Large Reasoning Models

Recent advances in Large Language Models have led to Large Reasoning Models, which produce step-by-step reasoning traces. These traces offer insight into how models think and their goals, improving explainability and helping users follow the logic, learn the process, and even debug errors. These traces, however, are often verbose and complex, making them cognitively demanding to comprehend. We address this challenge with ReTrace, an interactive system that structures and visualizes textual reasoning traces to support understanding. We use a validated reasoning taxonomy to produce structured reasoning data and investigate two types of interactive visualizations thereof. In a controlled user study, both visualizations enabled users to comprehend the model's reasoning more accurately and with less perceived effort than a raw text baseline. The results of this study could have design implications for making long and complex machine-generated reasoning processes more usable and transparent, an important step in AI explainability.

cs.HC↗

arXiv2025

Vision-Based Mobile App GUI Testing: A Survey

Graphical User Interface (GUI) has become one of the most significant parts of mobile applications (apps). It is a direct bridge between mobile apps and end users, which directly affects the end user's experience. Neglecting GUI quality can undermine the value and effectiveness of the entire mobile app solution. Significant research efforts have been devoted to GUI testing, one effective method to ensure mobile app quality. By conducting rigorous GUI testing, developers can ensure that the visual and interactive elements of the mobile apps not only meet functional requirements but also provide a seamless and user-friendly experience. However, traditional solutions, relying on the source code or layout files, have met challenges in both effectiveness and efficiency due to the gap between what is obtained and what app GUI actually presents. Vision-based mobile app GUI testing approaches emerged with the development of computer vision technologies and have achieved promising progress. In this survey paper, we provide a comprehensive investigation of the state-of-the-art techniques on 271 papers, among which 92 are vision-based studies. This survey covers different topics of GUI testing, like GUI test generation, GUI test record & replay, GUI testing framework, etc. Specifically, the research emphasis of this survey is placed mostly on how vision-based techniques outperform traditional solutions and have gradually taken a vital place in the GUI testing field. Based on the investigation of existing studies, we outline the challenges and opportunities of (vision-based) mobile app GUI testing and propose promising research directions with the combination of emerging techniques.

cs.SE↗

arXiv2025

Breaking Single-Tester Limits: Multi-Agent LLMs for Multi-User Feature Testing

The growing dependence on mobile phones and their apps has made multi-user interactive features, like chat calls, live streaming, and video conferencing, indispensable for bridging the gaps in social connectivity caused by physical and situational barriers. However, automating these interactive features for testing is fraught with challenges, owing to their inherent need for timely, dynamic, and collaborative user interactions, which current automated testing methods inadequately address. Inspired by the concept of agents designed to autonomously and collaboratively tackle problems, we propose MAdroid, a novel multi-agent approach powered by the Large Language Models (LLMs) to automate the multi-user interactive task for app feature testing. Specifically, MAdroid employs two functional types of multi-agents: user agents (Operator) and supervisor agents (Coordinator and Observer). Each agent takes a specific role: the Coordinator directs the interactive task; the Operator mimics user interactions on the device; and the Observer monitors and reviews the task automation process. Our evaluation, which included 41 multi-user interactive tasks, demonstrates the effectiveness of our approach, achieving 82.9% of the tasks with 96.8% action similarity, outperforming the ablation studies and state-of-the-art baselines. Additionally, a preliminary investigation underscores MAdroid's practicality by helping identify 11 multi-user interactive bugs during regression app testing, confirming its potential value in real-world software development contexts.

cs.SE↗

arXiv2025

Repair Ingredients Are All You Need: Improving Large Language Model-Based Program Repair via Repair Ingredients Search

Automated Program Repair (APR) techniques aim to automatically fix buggy programs. Among these, Large Language Model-based (LLM-based) approaches have shown great promise. Recent advances demonstrate that directly leveraging LLMs can achieve leading results. However, these techniques remain suboptimal in generating contextually relevant and accurate patches, as they often overlook repair ingredients crucial for practical program repair. In this paper, we propose ReinFix, a novel framework that enables LLMs to autonomously search for repair ingredients throughout both the reasoning and solution phases of bug fixing. In the reasoning phase, ReinFix integrates static analysis tools to retrieve internal ingredients, such as variable definitions, to assist the LLM in root cause analysis when it encounters difficulty understanding the context. During the solution phase, when the LLM lacks experience in fixing specific bugs, ReinFix searches for external ingredients from historical bug fixes with similar bug patterns, leveraging both the buggy code and its root cause to guide the LLM in identifying appropriate repair actions, thereby increasing the likelihood of generating correct patches. Evaluations on two popular benchmarks (Defects4J V1.2 and V2.0) demonstrate the effectiveness of our approach over SOTA baselines. Notably, ReinFix fixes 146 bugs, which is 32 more than the baselines on Defects4J V1.2. On Defects4J V2.0, ReinFix fixes 38 more bugs than the SOTA. Importantly, when evaluating on the recent benchmarks that are free of data leakage risk, ReinFix also maintains the best performance.

cs.SE↗

arXiv2025

Seeing is Fixing: Cross-Modal Reasoning with Multimodal LLMs for Visual Software Issue Fixing

Large language model-(LLM) based automated program repair (APR) techniques have shown promising results in resolving real-world GitHub issue tasks. Existing APR systems are primarily evaluated in unimodal settings (e.g., SWE-bench). However, these autonomous systems struggle to resolve multimodal problem scenarios (e.g., SWE-bench M) due to limitations in interpreting and leveraging visual information. In multimodal scenarios, LLMs need to rely on visual information in the graphical user interface (GUI) to understand bugs and generate fixes. To bridge this gap, we propose GUIRepair, a cross-modal reasoning approach for resolving multimodal issue scenarios by understanding and capturing visual information. Specifically, GUIRepair integrates two key components, Image2Code and Code2Image, to enhance fault comprehension and patch validation. Image2Code extracts relevant project documents based on the issue report, then applies this domain knowledge to generate the reproduced code responsible for the visual symptoms, effectively translating GUI images into executable context for better fault comprehension. Code2Image replays the visual issue scenario using the reproduced code and captures GUI renderings of the patched program to assess whether the fix visually resolves the issue, providing feedback for patch validation. We evaluate GUIRepair on SWE-bench M, and the approach demonstrates significant effectiveness. When utilizing GPT-4o as the base model, GUIRepair solves 157 instances, outperforming the best open-source baseline by 26 instances. Furthermore, when using o4-mini as the base model, GUIRepair can achieve even better results and solve 175 instances, outperforming the top commercial system by 22 instances. This emphasizes the success of our new perspective on incorporating cross-modal reasoning by understanding and capturing visual information to resolve multimodal issues.

cs.SE↗

arXiv2025

A Semantic-Aware Layer-Freezing Approach to Computation-Efficient Fine-Tuning of Language Models

Finetuning language models (LMs) is crucial for adapting the models to downstream data and tasks. However, full finetuning is usually costly. Existing work, such as parameter-efficient finetuning (PEFT), often focuses on \textit{how to finetune} but neglects the issue of \textit{where to finetune}. As a pioneering work on reducing the cost of backpropagation (at the layer level) by answering where to finetune, we conduct a semantic analysis of the LM inference process. We first propose using transition traces of the latent representation to compute deviations (or loss). Then, using a derived formula of scaling law, we estimate the gain of each layer in reducing deviation (or loss). Further, we narrow down the scope for finetuning, and also, study the cost-benefit balance of LM finetuning. We perform extensive experiments across well-known LMs and datasets. The results show that our approach is effective and efficient, and outperforms the existing baselines. Our approach is orthogonal to other techniques for improving finetuning efficiency, such as PEFT methods, offering practical values on LM finetuning.

cs.CL↗

arXiv2025

SeMe: Training-Free Language Model Merging via Semantic Alignment

Despite the remarkable capabilities of Language Models (LMs) across diverse tasks, no single model consistently outperforms others, necessitating efficient methods to combine their strengths without expensive retraining. Existing model merging techniques, such as parameter averaging and task-guided fusion, often rely on data-dependent computations or fail to preserve internal knowledge, limiting their robustness and scalability. We introduce SeMe (Semantic-based Merging), a novel, data-free, and training-free approach that leverages latent semantic alignment to merge LMs at a fine-grained, layer-wise level. Unlike prior work, SeMe not only preserves model behaviors but also explicitly stabilizes internal knowledge, addressing a critical gap in LM fusion. Through extensive experiments across diverse architectures and tasks, we demonstrate that SeMe outperforms existing methods in both performance and efficiency while eliminating reliance on external data. Our work establishes a new paradigm for knowledge-aware model merging and provides insights into the semantic structure of LMs, paving the way for more scalable and interpretable model composition.

cs.CL↗

arXiv2025

ai.txt: A Domain-Specific Language for Guiding AI Interactions with the Internet

We introduce ai.txt, a novel domain-specific language (DSL) designed to explicitly regulate interactions between AI models, agents, and web content, addressing critical limitations of the widely adopted robots.txt standard. As AI increasingly engages with online materials for tasks such as training, summarization, and content modification, existing regulatory methods lack the necessary granularity and semantic expressiveness to ensure ethical and legal compliance. ai.txt extends traditional URL-based access controls by enabling precise element-level regulations and incorporating natural language instructions interpretable by AI systems. To facilitate practical deployment, we provide an integrated development environment with code autocompletion and automatic XML generation. Furthermore, we propose two compliance mechanisms: XML-based programmatic enforcement and natural language prompt integration, and demonstrate their effectiveness through preliminary experiments and case studies. Our approach aims to aid the governance of AI-Internet interactions, promoting responsible AI use in digital ecosystems.

cs.NI↗

arXiv2025

Agent for User: Testing Multi-User Interactive Features in TikTok

TikTok, a widely-used social media app boasting over a billion monthly active users, requires effective app quality assurance for its intricate features. Feature testing is crucial in achieving this goal. However, the multi-user interactive features within the app, such as live streaming, voice calls, etc., pose significant challenges for developers, who must handle simultaneous device management and user interaction coordination. To address this, we introduce a novel multi-agent approach, powered by the Large Language Models (LLMs), to automate the testing of multi-user interactive app features. In detail, we build a virtual device farm that allocates the necessary number of devices for a given multi-user interactive task. For each device, we deploy an LLM-based agent that simulates a user, thereby mimicking user interactions to collaboratively automate the testing process. The evaluations on 24 multi-user interactive tasks within the TikTok app, showcase its capability to cover 75% of tasks with 85.9% action similarity and offer 87% time savings for developers. Additionally, we have also integrated our approach into the real-world TikTok testing platform, aiding in the detection of 26 multi-user interactive bugs.

cs.SE↗

arXiv2025

From Prompts to Templates: A Systematic Prompt Template Analysis for Real-world LLMapps

Large Language Models (LLMs) have revolutionized human-AI interaction by enabling intuitive task execution through natural language prompts. Despite their potential, designing effective prompts remains a significant challenge, as small variations in structure or wording can result in substantial differences in output. To address these challenges, LLM-powered applications (LLMapps) rely on prompt templates to simplify interactions, enhance usability, and support specialized tasks such as document analysis, creative content generation, and code synthesis. However, current practices heavily depend on individual expertise and iterative trial-and-error processes, underscoring the need for systematic methods to optimize prompt template design in LLMapps. This paper presents a comprehensive analysis of prompt templates in practical LLMapps. We construct a dataset of real-world templates from open-source LLMapps, including those from leading companies like Uber and Microsoft. Through a combination of LLM-driven analysis and human review, we categorize template components and placeholders, analyze their distributions, and identify frequent co-occurrence patterns. Additionally, we evaluate the impact of identified patterns on LLMs' instruction-following performance through sample testing. Our findings provide practical insights on prompt template design for developers, supporting the broader adoption and optimization of LLMapps in industrial settings.

cs.SE↗

arXiv2025

THEMIS: Towards Practical Intellectual Property Protection for Post-Deployment On-Device Deep Learning Models

On-device deep learning (DL) has rapidly gained adoption in mobile apps, offering the benefits of offline model inference and user privacy preservation over cloud-based approaches. However, it inevitably stores models on user devices, introducing new vulnerabilities, particularly model-stealing attacks and intellectual property infringement. While system-level protections like Trusted Execution Environments (TEEs) provide a robust solution, practical challenges remain in achieving scalable on-device DL model protection, including complexities in supporting third-party models and limited adoption in current mobile solutions. Advancements in TEE-enabled hardware, such as NVIDIA's GPU-based TEEs, may address these obstacles in the future. Currently, watermarking serves as a common defense against model theft but also faces challenges here as many mobile app developers lack corresponding machine learning expertise and the inherent read-only and inference-only nature of on-device DL models prevents third parties like app stores from implementing existing watermarking techniques in post-deployment models. To protect the intellectual property of on-device DL models, in this paper, we propose THEMIS, an automatic tool that lifts the read-only restriction of on-device DL models by reconstructing their writable counterparts and leverages the untrainable nature of on-device DL models to solve watermark parameters and protect the model owner's intellectual property. Extensive experimental results across various datasets and model structures show the superiority of THEMIS in terms of different metrics. Further, an empirical investigation of 403 real-world DL mobile apps from Google Play is performed with a success rate of 81.14%, showing the practicality of THEMIS.

cs.CR↗

arXiv2025

AccessFixer: Enhancing GUI Accessibility for Low Vision Users With R-GCN Model

The Graphical User Interface (GUI) plays a critical role in the interaction between users and mobile applications (apps), aiming at facilitating the operation process. However, due to the variety of functions and non-standardized design, GUIs might have many accessibility issues, like the size of components being too small or their intervals being narrow. These issues would hinder the operation of low vision users, preventing them from obtaining information accurately and conveniently. Although several technologies and methods have been proposed to address these issues, they are typically confined to issue identification, leaving the resolution in the hands of developers. Moreover, it can be challenging to ensure that the color, size, and interval of the fixed GUIs are appropriately compared to the original ones. In this work, we propose a novel approach named AccessFixer, which utilizes the Relational-Graph Convolutional Neural Network (R-GCN) to simultaneously fix three kinds of accessibility issues, including small sizes, narrow intervals, and low color contrast in GUIs. With AccessFixer, the fixed GUIs would have a consistent color palette, uniform intervals, and adequate size changes achieved through coordinated adjustments to the attributes of related components. Our experiments demonstrate the effectiveness and usefulness of AccessFixer in fixing GUI accessibility issues. After fixing 30 real-world apps, our approach solves an average of 81.2% of their accessibility issues. Also, we apply AccessFixer to 10 open-source apps by submitting the fixed results with pull requests (PRs) on GitHub. The results demonstrate that developers approve of our submitted fixed GUIs, with 8 PRs being merged or under fixing. A user study examines that low vision users host a positive attitude toward the GUIs fixed by our method.

cs.SE↗