arXiv subjects

Thanh Le-Cong

Publications and source records attributed to Thanh Le-Cong.

At least 19 recordsLinked to original sources

arXiv2026

Athena: Vulnerability-Affected Library Identification via Knowledge Graph Completion

A single vulnerability in a widely used library can cascade through millions of dependent applications, yet more than half of vulnerability database entries contain missing or incorrect affected-library information. Existing automated approaches neglect the relational structure of vulnerability databases, treating identification as an isolated text retrieval problem. In this paper, we propose Athena, the first graph-based approach for vulnerability affected library identification. Athena models vulnerability databases as a knowledge graph and reformulates the identification problem as knowledge graph completion (KGC). It comprises three key modules: a Modeling module that constructs a security knowledge graph integrating CVEs, libraries, CWE weakness types, CPE products, and software ecosystems; a Completion module that applies a modular KGC backbone to predict missing affected libraries for a given CVE via link prediction; and a Re-ranking module that retrieves KGC candidates and rescores them using a fine-tuned LLM augmented with knowledge graph embeddings, jointly leveraging structural and textual information. Our experiments on VulLib demonstrate that Athena significantly outperforms four state-of-the-art baselines, achieving a 32% improvement in Avg. F1 over the best baseline (i.e., VulLibGen). Notably, our KGC backbone with only 110M parameters already surpasses VulLibGen's best configuration at 7B parameters, demonstrating the effectiveness of graph-based modeling; the re-ranking module then provides substantial further gains, consistently outperforming the best baseline across all evaluated LLM backbones.

cs.SE↗

arXiv2026

FPMoE: A Sparse Mixture-of-Experts Approach to Functional Code Generation

Despite rapid progress in LLM-based code generation, existing models are predominantly trained on imperative languages, leaving functional programming languages (FPLs) such as Haskell, OCaml, and Scala chronically underexplored, with even frontier models performing substantially worse on FPLs. Fine-tuning is a natural remedy, but our experiments show that per-language fine-tuning fails to capture shared functional abstractions, while merged multi-language fine-tuning introduces cross-language interference. To address this, we introduce FPMoE, a lightweight, open-source code generation model built on a sparse Mixture-of-Experts (MoE) architecture with three language-specific routed experts (one each for Haskell, OCaml, and Scala) and a shared expert that captures cross-language functional patterns such as monadic reasoning and type-directed programming. This design resolves both failure modes simultaneously: dedicated experts eliminate interference, while the shared expert preserves abstractions that per-language models miss. On FPEval, FPMoE substantially outperforms fine-tuned baselines and, with only 3B active parameters, matches the performance of much larger models including DeepSeek-Coder-6.7B, Qwen2.5-Coder-14B-Instruct, and Qwen3-Coder-30B-A3B.

cs.PL↗

arXiv2026

PatchGuru: Patch Oracle Inference from Natural Language Artifacts

As software systems evolve, patches may unintentionally alter program behavior. Validating patches against their intended semantics is difficult due to incomplete regression tests and informal, non-executable natural language (NL) descriptions of patch intent. We present PatchGuru, the first automated technique that infers executable patch specifications from real-world pull requests (PRs). Given a PR, PatchGuru uses large language models (LLMs) to extract developer intent from NL artifacts and synthesizes patch oracles: under-approximate yet practical specifications expressed as runtime assertions in comparison programs that integrate pre- and post-patch versions. Patch oracles focus on patch-relevant behaviors, enable automated validation, and support cross-version properties. PatchGuru iteratively refines inferred oracles by comparing pre- and post-patch behaviors, identifies violations, filters inconsistencies via self-review, and generates bug reports. We evaluate PatchGuru on 400 recent PRs from four widely used open-source Python projects. PatchGuru reports 39 warnings with a precision of 0.62, yielding 24 confirmed true positives, including 12 previously unknown bugs, 11 of which were subsequently fixed by developers. Compared to the state-of-the-art technique Testora, PatchGuru detects 17 more bugs (24 vs. 7) while improving precision from 0.32 to 0.62. PatchGuru incurs an average cost of 8.9 minutes and USD 0.07 per PR. These results suggest that PatchGuru complements code review and regression testing by providing executable documentation and automated validation of patch intent.

cs.SE↗

arXiv2026

Perish or Flourish? A Holistic Evaluation of Large Language Models for Code Generation in Functional Programming

Functional programming provides strong foundations for developing reliable and secure software systems, yet its adoption remains not widespread due to the steep learning curve. Recent advances in Large Language Models (LLMs) for code generation present new opportunities to lower these barriers. However, extensive evaluations of LLMs largely focus on imperative programming languages, and their capabilities in functional programming languages (FP) remain underexplored. To address this gap, we introduce FPEval, a holistic evaluation framework built on FPBench, a new benchmark of 721 programming tasks across three difficulty levels on three mainstream FP languages: Haskell, Ocaml and Scala. FPEval provides compehensive evaluation infrastructures with both test validations with comprehensive test suites and static analysis tools to assess both functional correctness and code style and maintainability. Using this framework, we evaluate state-of-the-art LLMs, including GPT-3.5, GPT-4o, and GPT-5, for code generation in functional programming languages and Java as an imperative baseline. Our results demonstrate that LLM performance in functional programming improves substantially with model advancement; however, error rates remain significantly higher in purely functional languages (Haskell and OCaml) than in hybrid (Scala) or imperative (Java) languages. Moreover, LLMs frequently generate non-idiomatic functional code that follows imperative patterns, raising concerns about code style and long-term maintainability. Finally, we show that LLMs can partially self-repair both correctness and quality issues when provided with static analysis feedback and hand-crafted instructions for common types of issues.

cs.PL↗

arXiv2025

How Do Semantically Equivalent Code Transformations Impact Membership Inference on LLMs for Code?

The success of large language models for code relies on vast amounts of code data, including public open-source repositories, such as GitHub, and private, confidential code from companies. This raises concerns about intellectual property compliance and the potential unauthorized use of license-restricted code. While membership inference (MI) techniques have been proposed to detect such unauthorized usage, their effectiveness can be undermined by semantically equivalent code transformation techniques, which modify code syntax while preserving semantic. In this work, we systematically investigate whether semantically equivalent code transformation rules might be leveraged to evade MI detection. The results reveal that model accuracy drops by only 1.5% in the worst case for each rule, demonstrating that transformed datasets can effectively serve as substitutes for fine-tuning. Additionally, we find that one of the rules (RenameVariable) reduces MI success by 10.19%, highlighting its potential to obscure the presence of restricted code. To validate these findings, we conduct a causal analysis confirming that variable renaming has the strongest causal effect in disrupting MI detection. Notably, we find that combining multiple transformations does not further reduce MI effectiveness. Our results expose a critical loophole in license compliance enforcement for training large language models for code, showing that MI detection can be substantially weakened by transformation-based obfuscation techniques.

cs.SE↗

arXiv2025

Signature in Code Backdoor Detection, how far are we?

As Large Language Models (LLMs) become increasingly integrated into software development workflows, they also become prime targets for adversarial attacks. Among these, backdoor attacks are a significant threat, allowing attackers to manipulate model outputs through hidden triggers embedded in training data. Detecting such backdoors remains a challenge, and one promising approach is the use of Spectral Signature defense methods that identify poisoned data by analyzing feature representations through eigenvectors. While some prior works have explored Spectral Signatures for backdoor detection in neural networks, recent studies suggest that these methods may not be optimally effective for code models. In this paper, we revisit the applicability of Spectral Signature-based defenses in the context of backdoor attacks on code models. We systematically evaluate their effectiveness under various attack scenarios and defense configurations, analyzing their strengths and limitations. We found that the widely used setting of Spectral Signature in code backdoor detection is often suboptimal. Hence, we explored the impact of different settings of the key factors. We discovered a new proxy metric that can more accurately estimate the actual performance of Spectral Signature without model retraining after the defense.

cs.SE↗

arXiv2025

PatchSeeker: Mapping NVD Records to their Vulnerability-fixing Commits with LLM Generated Commits and Embeddings

Software vulnerabilities pose serious risks to modern software ecosystems. While the National Vulnerability Database (NVD) is the authoritative source for cataloging these vulnerabilities, it often lacks explicit links to the corresponding Vulnerability-Fixing Commits (VFCs). VFCs encode precise code changes, enabling vulnerability localization, patch analysis, and dataset construction. Automatically mapping NVD records to their true VFCs is therefore critical. Existing approaches have limitations as they rely on sparse, often noisy commit messages and fail to capture the deep semantics in the vulnerability descriptions. To address this gap, we introduce PatchSeeker, a novel method that leverages large language models to create rich semantic links between vulnerability descriptions and their VFCs. PatchSeeker generates embeddings from NVD descriptions and enhances commit messages by synthesizing detailed summaries for those that are short or uninformative. These generated messages act as a semantic bridge, effectively closing the information gap between natural language reports and low-level code changes. Our approach PatchSeeker achieves 59.3% higher MRR and 27.9% higher Recall@10 than the best-performing baseline, Prospector, on the benchmark dataset. The extended evaluation on recent CVEs further confirms PatchSeeker's effectiveness. Ablation study shows that both the commit message generation method and the selection of backbone LLMs make a positive contribution to PatchSeeker. We also discuss limitations and open challenges to guide future work.

cs.SE↗

arXiv2025

VulGuard: An Unified Tool for Evaluating Just-In-Time Vulnerability Prediction Models

We present VulGuard, an automated tool designed to streamline the extraction, processing, and analysis of commits from GitHub repositories for Just-In-Time vulnerability prediction (JIT-VP) research. VulGuard automatically mines commit histories, extracts fine-grained code changes, commit messages, and software engineering metrics, and formats them for downstream analysis. In addition, it integrates several state-of-the-art vulnerability prediction models, allowing researchers to train, evaluate, and compare models with minimal setup. By supporting both repository-scale mining and model-level experimentation within a unified framework, VulGuard addresses key challenges in reproducibility and scalability in software security research. VulGuard can also be easily integrated into the CI/CD pipeline. We demonstrate the effectiveness of the tool in two influential open-source projects, FFmpeg and the Linux kernel, highlighting its potential to accelerate real-world JIT-VP research and promote standardized benchmarking. A demo video is available at: https://youtu.be/j96096-pxbs

cs.SE↗

arXiv2025

Toward Realistic Evaluations of Just-In-Time Vulnerability Prediction

Modern software systems are increasingly complex, presenting significant challenges in quality assurance. Just-in-time vulnerability prediction (JIT-VP) is a proactive approach to identifying vulnerable commits and providing early warnings about potential security risks. However, we observe that current JIT-VP evaluations rely on an idealized setting, where the evaluation datasets are artificially balanced, consisting exclusively of vulnerability-introducing and vulnerability-fixing commits. To address this limitation, this study assesses the effectiveness of JIT-VP techniques under a more realistic setting that includes both vulnerability-related and vulnerability-neutral commits. To enable a reliable evaluation, we introduce a large-scale public dataset comprising over one million commits from FFmpeg and the Linux kernel. Our empirical analysis of eight state-of-the-art JIT-VP techniques reveals a significant decline in predictive performance when applied to real-world conditions; for example, the average PR-AUC on Linux drops 98% from 0.805 to 0.016. This discrepancy is mainly attributed to the severe class imbalance in real-world datasets, where vulnerability-introducing commits constitute only a small fraction of all commits. To mitigate this issue, we explore the effectiveness of widely adopted techniques for handling dataset imbalance, including customized loss functions, oversampling, and undersampling. Surprisingly, our experimental results indicate that these techniques are ineffective in addressing the imbalance problem in JIT-VP. These findings underscore the importance of realistic evaluations of JIT-VP and the need for domain-specific techniques to address data imbalance in such scenarios.

cs.SE↗

arXiv2025

From Empirical Evaluation to Context-Aware Enhancement: Repairing Regression Errors with LLMs

[...] Since then, various APR approaches, especially those leveraging the power of large language models (LLMs), have been rapidly developed to fix general software bugs. Unfortunately, the effectiveness of these advanced techniques in the context of regression bugs remains largely unexplored. This gap motivates the need for an empirical study evaluating the effectiveness of modern APR techniques in fixing real-world regression bugs. In this work, we conduct an empirical study of APR techniques on regression bugs. To facilitate our study, we introduce RegressionBug4APR, a high-quality benchmark of Java and Python regression bugs integrated into a framework designed to facilitate APR research. The current benchmark includes 200 regression bugs collected from widely used real-world GitHub repositories. We begin by conducting an in-depth analysis of the benchmark, demonstrating its diversity and quality. Building on this foundation, we empirically evaluate the capabilities of APR to regression bugs by assessing both traditional APR tools and advanced LLM-based APR approaches. Our experimental results show that classical APR tools fail to repair any bugs, while LLM-based APR approaches exhibit promising potential. Motivated by these results, we investigate impact of incorporating bug-inducing change information into LLM-based APR approaches for fixing regression bugs. We further conduct an ablation study to disaggregate the contribution of each contextual element within the bug-inducing change information. Our results highlight that this context-aware enhancement significantly improves the performance of LLM-based APR, yielding 1.6x more successful repairs compared to using LLM-based APR without such context. Moreover, our findings are consistent across both Java and Python benchmarks, providing preliminary evidence for the generalizability of our findings.

cs.SE↗

arXiv2025

Mapping NVD Records to Their Vulnerability-fixing Commits: How Hard is It?

Mapping National Vulnerability Database (NVD) records to vulnerability-fixing commits (VFCs) is crucial for vulnerability analysis but challenging due to sparse explicit links in NVD references. This study explores this mapping's feasibility through an empirical approach. Manual analysis of NVD references showed Git references enable over 86% success, while non-Git references achieve under 14%. Using these findings, we built an automated pipeline extracting 31,942 VFCs from 20,360 NVD records (8.7% of 235,341) with 87% precision, mainly from Git references. To fill gaps, we mined six external security databases, yielding 29,254 VFCs for 18,985 records (8.1%) at 88.4% precision, and GitHub repositories, adding 3,686 VFCs for 2,795 records (1.2%) at 73% precision. Combining these, we mapped 26,710 unique records (11.3% coverage) from 7,634 projects, with overlap between NVD and external databases, plus unique GitHub contributions. Despite success with Git references, 88.7% of records remain unmapped, highlighting the difficulty without Git links. This study offers insights for enhancing vulnerability datasets and guiding future automated security research.

cs.SE↗

arXiv2025

Can LLMs Reason About Program Semantics? A Comprehensive Evaluation of LLMs on Formal Specification Inference

Large Language Models (LLMs) are increasingly being used to automate programming tasks. Yet, LLMs' capabilities in reasoning about program semantics are still inadequately studied, leaving significant potential for further exploration. This paper introduces FormalBench, a comprehensive benchmark designed to evaluate LLMs' reasoning abilities on program semantics, particularly via the task of synthesizing formal program specifications to assist verifying program correctness. This task requires both comprehensive reasoning over all possible program executions and the generation of precise, syntactically correct expressions that adhere to formal syntax and semantics. Using this benchmark, we evaluated the ability of LLMs in synthesizing consistent and complete specifications. Our findings show that LLMs perform well with simple control flows but struggle with more complex structures, especially loops, even with advanced prompting. Additionally, LLMs exhibit limited robustness against semantic-preserving transformations. We also highlight common failure patterns and design self-repair prompts, improving success rates by 25%.

cs.PL↗

arXiv2024

Memory-Efficient Large Language Models for Program Repair with Semantic-Guided Patch Generation

In this paper, we first show that increases in beam size, even for small-sized LLMs (1B-7B params), require extensive GPU usage, leading to up to 80% of recurring crashes due to memory overloads in LLM-based APR. Seemingly simple solutions to reduce memory consumption are (1) to quantize LLM models, i.e., converting the weights of an LLM from high-precision values to lower-precision ones, and (2) to make beam search sequential, i.e., forwarding each beam through the model sequentially and then concatenating them back into a single output. However, we show that these approaches still do not work via both theoretical analysis and experiments. To address this, we introduce FLAMES, a novel LLM-based APR technique that employs semantic-guided patch generation to enhance repair effectiveness and memory efficiency. Unlike conventional methods that rely on beam search, FLAMES utilizes greedy decoding to enhance memory efficiency while steering the search towards more potentially good repair candidates via a semantic-guided best-first search algorithm. At each decoding step, FLAMES uses semantic feedback from test validation, such as the number of passing and failing test cases, to select the most promising token to explore further. Our empirical evaluation on Defects4J shows thatFLAMES substantially reduces memory consumption by up to 83% compared to LLM-based APR without compromising time efficiency. Moreover, FLAMES correctly fixes 133 bugs on Defects4J, fixing 10 bugs more than the best baseline. Additionally, these improvements also generalize to the HumanEval-Java and TransformedD4J datasets, where FLAMES generates 12% and 36.5% more correct patches, respectively, than the best baseline.

cs.SE↗

arXiv2024

Comparison of Static Application Security Testing Tools and Large Language Models for Repo-level Vulnerability Detection

Software vulnerabilities pose significant security challenges and potential risks to society, necessitating extensive efforts in automated vulnerability detection. There are two popular lines of work to address automated vulnerability detection. On one hand, Static Application Security Testing (SAST) is usually utilized to scan source code for security vulnerabilities, especially in industries. On the other hand, deep learning (DL)-based methods, especially since the introduction of large language models (LLMs), have demonstrated their potential in software vulnerability detection. However, there is no comparative study between SAST tools and LLMs, aiming to determine their effectiveness in vulnerability detection, understand the pros and cons of both SAST and LLMs, and explore the potential combination of these two families of approaches. In this paper, we compared 15 diverse SAST tools with 12 popular or state-of-the-art open-source LLMs in detecting software vulnerabilities from repositories of three popular programming languages: Java, C, and Python. The experimental results showed that SAST tools obtain low vulnerability detection rates with relatively low false positives, while LLMs can detect up 90\% to 100\% of vulnerabilities but suffer from high false positives. By further ensembling the SAST tools and LLMs, the drawbacks of both SAST tools and LLMs can be mitigated to some extent. Our analysis sheds light on both the current progress and future directions for software vulnerability detection.

cs.SE↗

arXiv2024

Towards Reliable Evaluation of Neural Program Repair with Natural Robustness Testing

In this paper, we propose shifting the focus of robustness evaluation for Neural Program Repair (NPR) techniques toward naturally-occurring data transformations. To accomplish this, we first examine the naturalness of semantic-preserving transformations through a two-stage human study. This study includes (1) interviews with senior software developers to establish concrete criteria for evaluating the naturalness of these transformations, and (2) a survey involving 10 developers to assess the naturalness of 1,178 transformations, i.e., pairs of original and transformed programs, applied to 225 real-world bugs. Our findings show that only 60% of these transformations are deemed natural, while 20% are considered unnatural, with strong agreement among annotators. Moreover, the unnaturalness of these transformations significantly impacts both their applicability to benchmarks and the conclusions drawn from robustness testing. Next, we conduct natural robustness testing on NPR techniques to assess their true effectiveness against real-world data variations. Our experimental results reveal a substantial number of prediction changes in NPR techniques, leading to significant reductions in both plausible and correct patch rates when comparing performance on the original and transformed datasets. Additionally, we observe notable differences in performance improvements between NPR techniques, suggesting potential biases on NPR evaluation introduced by limited datasets. Finally, we propose an LLM-based metric to automate the assessment of transformation naturalness, ensuring the scalability of natural robustness testing.

cs.SE↗

arXiv2023

Refining ChatGPT-Generated Code: Characterizing and Mitigating Code Quality Issues

We systematically study the quality of 4,066 ChatGPT-generated code implemented in two popular programming languages, i.e., Java and Python, for 2,033 programming tasks. The goal of this work is three folds. First, we analyze the correctness of ChatGPT on code generation tasks and uncover the factors that influence its effectiveness, including task difficulty, programming language, time that tasks are introduced, and program size. Second, we identify and characterize potential issues with the quality of ChatGPT-generated code. Last, we provide insights into how these issues can be mitigated. Experiments highlight that out of 4,066 programs generated by ChatGPT, 2,756 programs are deemed correct, 1,082 programs provide wrong outputs, and 177 programs contain compilation or runtime errors. Additionally, we further analyze other characteristics of the generated code through static analysis tools, such as code style and maintainability, and find that 1,930 ChatGPT-generated code snippets suffer from maintainability issues. Subsequently, we investigate ChatGPT's self-repairing ability and its interaction with static analysis tools to fix the errors uncovered in the previous step. Experiments suggest that ChatGPT can partially address these challenges, improving code quality by more than 20%, but there are still limitations and opportunities for improvement. Overall, our study provides valuable insights into the current limitations of ChatGPT and offers a roadmap for future research and development efforts to enhance the code generation capabilities of AI models like ChatGPT.

cs.SE↗

arXiv2023

Are We Ready to Embrace Generative AI for Software Q&A?

Stack Overflow, the world's largest software Q&A (SQA) website, is facing a significant traffic drop due to the emergence of generative AI techniques. ChatGPT is banned by Stack Overflow after only 6 days from its release. The main reason provided by the official Stack Overflow is that the answers generated by ChatGPT are of low quality. To verify this, we conduct a comparative evaluation of human-written and ChatGPT-generated answers. Our methodology employs both automatic comparison and a manual study. Our results suggest that human-written and ChatGPT-generated answers are semantically similar, however, human-written answers outperform ChatGPT-generated ones consistently across multiple aspects, specifically by 10% on the overall score. We release the data, analysis scripts, and detailed results at https://anonymous.4open.science/r/GAI4SQA-FD5C.

cs.SE↗

arXiv2023

Multi-Granularity Detector for Vulnerability Fixes

With the increasing reliance on Open Source Software, users are exposed to third-party library vulnerabilities. Software Composition Analysis (SCA) tools have been created to alert users of such vulnerabilities. SCA requires the identification of vulnerability-fixing commits. Prior works have proposed methods that can automatically identify such vulnerability-fixing commits. However, identifying such commits is highly challenging, as only a very small minority of commits are vulnerability fixing. Moreover, code changes can be noisy and difficult to analyze. We observe that noise can occur at different levels of detail, making it challenging to detect vulnerability fixes accurately. To address these challenges and boost the effectiveness of prior works, we propose MiDas (Multi-Granularity Detector for Vulnerability Fixes). Unique from prior works, Midas constructs different neural networks for each level of code change granularity, corresponding to commit-level, file-level, hunk-level, and line-level, following their natural organization. It then utilizes an ensemble model that combines all base models to generate the final prediction. This design allows MiDas to better handle the noisy and highly imbalanced nature of vulnerability-fixing commit data. Additionally, to reduce the human effort required to inspect code changes, we have designed an effort-aware adjustment for Midas's outputs based on commit length. The evaluation results demonstrate that MiDas outperforms the current state-of-the-art baseline in terms of AUC by 4.9% and 13.7% on Java and Python-based datasets, respectively. Furthermore, in terms of two effort-aware metrics, EffortCost@L and Popt@L, MiDas also outperforms the state-of-the-art baseline, achieving improvements of up to 28.2% and 15.9% on Java, and 60% and 51.4% on Python, respectively.

cs.CR↗