Enhanced Web Testing with LLMs: A Research Roadmap

With the rapid advancement of Large Language Models (LLMs), traditional web testing research faces increasingly severe challenges and higher demands. To advance web testing toward higher levels of intelligence and automation, it is essential to systematically investigate the application and development of LLMs in this field. Accordingly, this paper proposes a comprehensive research roadmap to guide these efforts. We structure this roadmap around four pivotal research directions that span the entire testing process: (1) Adapting LLMs for Web Testing (Pre-Testing); (2) The Role of LLMs in Web Testing (In-Testing); (3) Results Analysis and Decision Support (Post-Testing); and (4) Necessity of Using LLMs in Web Testing. We contend that rigorous exploration within these domains is critical for realizing the next generation of automated and intelligent web testing frameworks. For each research direction, we summarize the latest progress in LLMs applications for web testing and identify the challenges and remaining gaps that future research must address. While this roadmap is not exhaustive, our objective is to catalyze further inquiry within the academic community, thereby advancing the state-of-the-art and fully leveraging the potential of LLMs in web testing.