Comparison of Datasets Used in Research on Large Language Models for Requirements Engineering

This ORKG comparison summarizes the findings of a systematic mapping study analyzing 62 publicly available datasets referenced in 41 scientific publications involving Large Language Models for Requirements Engineering (LLM4RE). Each dataset is characterized along descriptors such as artifact type, RE stage, task, domain, and language. The analysis shows that most datasets focus on requirements and user reviews, mainly targeting management and analysis stages through classification and traceability tasks. However, the study reveals significant research gaps, including the scarcity of datasets for elicitation and late-stage RE activities, limited multilingual resources, small dataset sizes, and incomplete licensing or metadata. The resulting public catalogue and ORKG comparison aim to enhance dataset discoverability, comparison, and reuse, laying the groundwork for a shared data infrastructure in LLM-based RE research.