The ethical PROMISE NFR dataset: data for ethical concern classification in software requirements

The Ethical PROMISE NFR dataset is an extension of the widely utilized PROMISE Non-Functional Requirements (NFR) dataset annotated using a multi-label ethical annotation framework comprising nine ethical concern categories: transparency, accountability, fairness, bias, privacy, accessibility, user consent, sustainability, and human autonomy. The dataset contains 3677 software requirement statements collected from the PROMISE repository and manually tagged using a concern-based ethical taxonomy considering responsible AI principles, the General Data Protection Regulation (GDPR) and the European Union AI Act. To assess annotation reliability, the dataset was independently reviewed by two annotators. Direct inter-annotator agreement between Annotator 1 and Annotator 2 yielded Cohen's κ = 0.6164 with 80.11% observed agreement. Against the reference annotations, Annotator 1 achieved κ = 0.9093 and Annotator 2 achieved κ = 0.5585, with an average agreement of 86.21%. The dataset is publicly available on IEEE DataPort and Mendeley Data and can support research in ethical requirements engineering, responsible AI, software requirements classification, explainable artificial intelligence, and natural language processing. The data may also be used to develop, train, and evaluate machine learning and deep learning models for ethical concern detection in software requirements.