SmartSE

The automation of software requirements analysis and test case generation remains one of the most pressing challenges in modern software engineering. Traditional approaches rely heavily on manual effort, which is time-consuming, error-prone, and inconsistent, especially as systems grow in complexity. This paper presents SmartSE, a novel AI-driven framework that leverages Large Language Models (LLMs), specifically fine-tuned variants of GPT-4 and LLaMA-3, to automate two critical phases of the software development lifecycle (SDLC): (1) natural language requirements analysis and formalization and (2) intelligent test case generation. The proposed framework integrates a multi-stage pipeline comprising requirement parsing, ambiguity detection, semantic enrichment, and test oracle synthesis. The authors evaluate SmartSE on three real-world open-source projects, Apache Kafka, Mozilla Firefox, and OpenMRS, spanning over 14,000 requirement statements and 21,000 existing test cases.