Towards Supporting Quality Architecture Evaluation with LLM Tools

This feed contains multiple software-engineering research papers exploring the application of LLMs, program analysis, and formal methods to improve development, debugging, testing, and verification. Key contributions include: using LLMs (MS Copilot, GPT-4) to assist architecture-quality evaluation and FSM synthesis; Lumos, a static-analysis-driven, low-overhead provenance framework for online distributed-system debugging; ReLog, an iterative LLM-guided logging generation pipeline optimized for downstream debugging/repair tasks; SemLoc, a structured semantic-grounding approach for LLM-aided,ver