FR_NFR_dataset

This dataset aims to support the requirement engineering phase of software development, in the early phase of requirement analysis of Functional Requirements (FR) and Non-Functional Requirements (NFR) for cross-cutting concerns and to improve modularity. The dataset will contain a collection of software requirements expressed in natural language.
Categories: Each requirement is categorized as either FR or NFR.
Subcategories (Optional): NFRs can be further categorized based on specific quality aspects like security, performance, usability, etc.
The dataset will highlight how certain requirements impact multiple modules or functionalities within the system.
The dataset can be used to train Natural Language Processing (NLP) tools to automatically analyze requirements annotated as FR or NFR.
Sources for the dataset are existing software requirements repositories like PROMISE https://arxiv.org/pdf/2211.05286 and open-source project documentation with FRs and NFRs. By processing existing datasets and containing open software project requirement documents we have a dataset of 6118 requirements where 3964 requirements are functional and 2154 requirements are nonfunctional However, creating a comprehensive dataset that covers a wide range of software systems and requirements is challenging as natural language requirements can be ambiguous.