论文研读 on Hello, I am Ruk1a.https://rukiaovo.github.io/categories/%E8%AE%BA%E6%96%87%E7%A0%94%E8%AF%BB/Recent content in 论文研读 on Hello, I am Ruk1a.Hugo -- gohugo.iozh-CNCopyright © 2026 Hello, I am Ruk1a.Sun, 05 Jul 2026 20:40:00 +0800AgentSpec: Customizable Runtime Enforcement for Safe and Reliable LLM Agentshttps://rukiaovo.github.io/posts/2026-07-01-agentspec-customizable-runtime-enforcement-for-safe-and-reliable-llm-agents/Sun, 05 Jul 2026 20:40:00 +0800https://rukiaovo.github.io/posts/2026-07-01-agentspec-customizable-runtime-enforcement-for-safe-and-reliable-llm-agents/<p>论文链接：https://arxiv.org/abs/2503.18666v3</p> <h2 class="relative group">论文主旨 <div id="论文主旨" class="anchor"></div> <span class="absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100 select-none"> <a class="text-primary-300 dark:text-neutral-700 !no-underline" href="#%e8%ae%ba%e6%96%87%e4%b8%bb%e6%97%a8" aria-label="锚点">#</a> </span> </h2> <p><strong>ICSE 2026，主要研究Agent运行时的安全约束，设计跨领域、易插拔的模块化框架用于规范Agent行为并降低风险。</strong></p>Self-Adaptation through Incremental Generative Model Transformations at Runtimehttps://rukiaovo.github.io/posts/2026-07-01-self-adaptation-through-incremental-generative-model-transformations-at-runtime/Fri, 03 Jul 2026 18:01:00 +0800https://rukiaovo.github.io/posts/2026-07-01-self-adaptation-through-incremental-generative-model-transformations-at-runtime/<p>论文链接：https://dl.acm.org/doi/abs/10.1145/2568225.2568310</p> <h2 class="relative group">论文主旨 <div id="论文主旨" class="anchor"></div> <span class="absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100 select-none"> <a class="text-primary-300 dark:text-neutral-700 !no-underline" href="#%e8%ae%ba%e6%96%87%e4%b8%bb%e6%97%a8" aria-label="锚点">#</a> </span> </h2> <p><strong>发表于2014 ICSE，主要研究self-adaptive system。</strong></p>Extracting Conceptual Knowledge to Locate Software Issueshttps://rukiaovo.github.io/posts/2026-07-01-extracting-conceptual-knowledge-to-locate-software-issues/Wed, 01 Jul 2026 22:12:00 +0800https://rukiaovo.github.io/posts/2026-07-01-extracting-conceptual-knowledge-to-locate-software-issues/<p>论文链接：https://arxiv.org/abs/2509.21427</p> <h2 class="relative group">论文主旨 <div id="论文主旨" class="anchor"></div> <span class="absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100 select-none"> <a class="text-primary-300 dark:text-neutral-700 !no-underline" href="#%e8%ae%ba%e6%96%87%e4%b8%bb%e6%97%a8" aria-label="锚点">#</a> </span> </h2> <p>主要研究如何增强LLM的根据bug/issue定位相关文件或代码的能力，提出了一种类似GraphRAG思想的issue localization增强方案。</p>