Exploring Large Language Models for Automated Non-Functional Requirements Generation

Exploring Large Language Models for Automated Non-Functional Requirements Generation: A Human Anotated Dataset for NFR Quality

This dataset is a resource designed to evaluvate the quality of NFRs generated by LLMs based solely on functional requirements.

This dataset included:

Human_Evaluvation_Data.tsv: A tab seperated file with human evaluation for the quality of NFRs and ISO/IEC 25010:2023 standard quality attributes, generated by LLMs

FR_34.tsv: Consisted of 34 functional requirements used to generate the dataset. This is a subset of the FR_NFR_Dataset.

AdvancedPrompt.txt has the prompt that we used to generate NFRs from FRs. It includes various prompt engineering techniques such as role assignment and knowledge grounded contraints to get LLMs to generate NFRs consistently 

For further details, please refer the linked paper.