RE-E2E

RE-E2E is a benchmark dataset for end-to-end software requirements engineering. It pairs concise project descriptions with human-written software requirements specifications (SRS) normalized to IEEE section structures.

Dataset contents

dataset.jsonl contains 47 records: 29 specifications normalized to IEEE 830-1998 and 18 normalized to IEEE 29148-2018. The IEEE 830-1998 records include their matching project summary. The IEEE 29148-2018 records have no matching project summaries in this release, so project_summary is null for those records.

Each record contains:

• project_id: source project identifier derived from the original filename.
• standard: the SRS section standard used for normalization.
• project_summary: project description text, or null when no matching summary is included.
• requirements: ordered section objects with section, subsection, subsubsection, and content fields. subsubsection is an empty string when the source has no such level.

Load it with the Datasets library:

from datasets import load_dataset

dataset = load_dataset("json", data_files="dataset.jsonl", split="train")

Source and citation

The dataset is archived on Zenodo as version 1: 10.5281/zenodo.22815961. The Agent4RE paper is available from the ACM Digital Library and arXiv. Please cite the paper:

@inproceedings{10.1145/3843779.3844634,
author = {Tang, Yongjian and Li, Linhan and Runkler, Thomas},
title = {Agent4RE: A Self-Refining Multi-agent Framework for End-to-End Software Requirements Engineering and Benchmarking},
year = {2026},
isbn = {9798400729911},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3843779.3844634},
doi = {10.1145/3843779.3844634},
abstract = {Existing LLM-based approaches for software Requirements Engineering (RE) typically rely on basic prompting strategies or rudimentary agent collaboration, under-utilizing the full potential of multi-agent systems. Meanwhile, available datasets focus on isolated subtasks, such as requirements extraction, classification, and completeness detection, leaving the absence of an end-to-end RE benchmark that spans from requirements elicitation to generation. We present Agent4RE - a self-refining multi-agent RE system that orchestrates specialized agents and incorporates two iterative improvement loops. To support evaluation, we construct RE-E2E - a real-world dataset built from human-written requirement specifications, enabling end-to-end assessment of RE workflows. Building on this foundation, we further propose two enhanced Agent4RE versions that incorporate either autonomous self-refinement or structured human feedback, and analyze their strengths and limitations across different scenarios. Evaluation on 8 Large Language Models (LLMs) demonstrates that all three Agent4RE variants consistently outperform a domain-context-augmented prompting baseline by average 8\% in text-based metrics. The two enhanced variants achieve the highest LLM-as-a-judge and human ratings, surpassing two RE baselines by approximately 0.8 points on a four-point scale. This consistent performance establishes Agent4RE as a practical end-to-end RE solution for industrial environments.},
booktitle = {Proceedings of the 1st International Workshop on PromptOps and Vibe Coding},
pages = {25–32},
numpages = {8},
keywords = {Large Language Models, Multi-agent Systems, Requirements Engineering, Software Engineering},
location = {Munich, Germany},
series = {POVC '26}
}

License

The dataset is distributed under the Creative Commons Attribution 4.0 International license.