{ "added_at": "2026-10-17", "arxiv_id": "2506.20170", "arxiv_version_reviewed": "v1", "authors": [ "Guoqiang Chen", "Xin Jin", "Zhiqiang Lin" ], "bibtex_url": "https://awesomerea.org/papers/citations/paper-jsdeobsbench.bib", "citation": { "container_title": "Proceedings of the 2025 ACM SIGSAC Conference on Computer and Communications Security", "pages": "36-50", "publisher": "ACM", "sources": [ "https://api.crossref.org/works/10.1145/3719027.3744871" ] }, "code_url": "https://github.com/Ch3nYe/JsDeObsBench", "date_basis": "ACM-deposited Crossref print-publication date; the October 13–17 CCS 2025 conference week is not used as the publication date.", "date_basis_zh": "采用 ACM 在 Crossref 登记的印刷出版日期；不将 CCS 2025 于 10 月 13—17 日举行的会议周作为出版日期。", "date_precision": "day", "description_en": "Builds an execution-verifiable JavaScript deobfuscation benchmark by applying seven JavaScript-Obfuscator transformations to Project CodeNet programs with test cases, then scores LLM and tool outputs for syntax, execution, code simplification and similarity to the original code.", "description_zh": "对附带测试用例的 Project CodeNet 程序施加 JavaScript-Obfuscator 的七种混淆变换，构建可通过执行验证的 JavaScript 反混淆基准，并从语法、执行结果、代码简化程度及与原始代码的相似度评估 LLM 和现有工具的输出。", "doi": "10.1145/3719027.3744871", "entry_url": "https://awesomerea.org/papers/paper-jsdeobsbench/", "entry_url_zh": "https://awesomerea.org/zh/papers/paper-jsdeobsbench/", "evidence_urls": [ "https://doi.org/10.1145/3719027.3744871", "https://api.crossref.org/works/10.1145/3719027.3744871", "https://arxiv.org/abs/2506.20170", "https://github.com/Ch3nYe/JsDeObsBench/blob/8b2e1c35dc4ae25a46ea580ea5dcb6b5de5edbe9/README.md", "https://github.com/Ch3nYe/JsDeObsBench/tree/8b2e1c35dc4ae25a46ea580ea5dcb6b5de5edbe9", "https://jsdeobf.github.io/" ], "first_arxiv_date": "2025-06-25", "id": "paper-jsdeobsbench", "json_url": "https://awesomerea.org/papers/paper-jsdeobsbench/record.json", "kind": "research", "markdown_url": "https://awesomerea.org/papers/paper-jsdeobsbench/index.md", "markdown_url_zh": "https://awesomerea.org/zh/papers/paper-jsdeobsbench/index.md", "metadata_notes": "ACM-deposited Crossref metadata records print publication on 2025-11-19 and online publication on 2025-11-22, pages 36–50. The reviewed text is arXiv v1, which carries the CCS 2025 header. The paper links the code repository and the jsdeobf.github.io leaderboard. At the pinned commit the repository contains the CodeNet-derived obfuscated datasets, evaluators and scripts, but no malware samples or license file.", "metadata_notes_zh": "ACM 在 Crossref 登记的元数据记载印刷出版于 2025-11-19，在线出版于 2025-11-22，页码为 36–50。审阅文本为带有 CCS 2025 页眉的 arXiv v1。论文给出了代码仓库和 jsdeobf.github.io 排行榜链接。固定提交中的仓库包含基于 CodeNet 生成的混淆数据集、评估器和脚本，但不含恶意代码样本，也没有许可证文件。", "pdf_url": "https://arxiv.org/pdf/2506.20170v1", "project_url": "https://jsdeobf.github.io/", "publication_date": "2025-11-19", "publication_status": "conference-paper", "record_type": "paper", "runtime_tested": false, "scope_limitations_en": "On average, LLM outputs passed execution checks 60.93% of the time, against 83% for the two existing deobfuscator baselines. All samples come from a single obfuscator applied to moderate-length competitive-programming solutions. The 4,515 obfuscated malware samples have no test cases and are compared only through behavior traces.", "scope_limitations_zh": "平均而言，LLM 输出通过执行检查的比例为 60.93%，两种现有反混淆工具基线为 83%。所有样本均由单一混淆器生成，原始程序是长度适中的编程竞赛解答。4,515 个经混淆的恶意代码样本没有测试用例，只能通过行为轨迹比较。", "targets": [ "website" ], "title": "JsDeObsBench: Measuring and Benchmarking LLMs for JavaScript Deobfuscation", "title_zh": "JsDeObsBench：衡量与评测大语言模型的 JavaScript 反混淆能力", "url": "https://doi.org/10.1145/3719027.3744871", "venue": "ACM CCS 2025", "verification": "source-reviewed", "verified_at": "2026-10-17", "why_fits_en": "Adds a measurable, execution-checked benchmark for LLM-assisted JavaScript deobfuscation, a common step in recovering website behavior from shipped scripts.", "why_fits_zh": "为 LLM 辅助的 JavaScript 反混淆提供可量化、经执行检验的基准；反混淆是从已部署脚本恢复网站行为的常见步骤。", "year": 2025 }