InteractiveReq: Enhancing Software Requirement Specification with Critiquing-based Recommender Systems

Specifying user stories and epics for feature requirements in agile software development is essential but time-consuming, demanding significant stakeholder effort. To improve this situation, we introduce InteractiveReq, an interactive critiquing-based recommender system that simplifies the generation of high-quality requirements. By using Large Language Models (LLMs), InteractiveReq enables an iterative process where stakeholders refine feature requirements through interactive feedback, addressing the issue of incomplete initial specifications. The system recommends custom drafts of epics and user stories based on the project context and an initial feature description, which users can refine through natural language critiques until their needs are satisfied. This approach aims to reduce workload and offers an intuitive method for requirement management. Preliminary results indicate that InteractiveReq effectively supports the creation of complete and accurate specifications.