Researchers have developed a method using a chain of LLMs with engineered prompts to automate the extraction of functional goals from software documentation. This approach involves actor identification and high/low-level goal extraction, incorporating a generation-critic mechanism as a feedback loop between two LLMs. While the pipeline achieved 61% accuracy in low-level goal identification, it is best suited to accelerate manual extraction rather than fully replace it. Future work aims to improve accuracy by integrating Retrieval-Augmented Generation (RAG) and Chain-of-Thought (CoT) prompting. AI
Summary written by gemini-2.5-flash-lite from 2 sources. How we write summaries →
IMPACT Provides a tool to accelerate manual goal extraction in software engineering, with potential for future improvements via RAG and CoT.
RANK_REASON Academic paper detailing a new method for LLM-based goal extraction in software engineering.