Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Jun 29:61:111842.
doi: 10.1016/j.dib.2025.111842. eCollection 2025 Aug.

NIT: A dataset for network intent translation

Ala' A AlSamarneh  1 , Omar Y Al-Jarrah  1 , Ahmad T Al-Hammouri  1   2
Affiliations

NIT: A dataset for network intent translation

Ala' A AlSamarneh et al. Data Brief. .

Abstract

This paper introduces the NIT (Network Intent Translations) dataset. The NIT dataset is designed for network intent translations, specifically for intent-to-vendor translations. The dataset contains network configuration scenarios described in natural language (intents) and their corresponding low-level configurations for Juniper EX3300 switches with JUNOS Base OS boot [12.3R12-S12]. The dataset was generated following quality measures by validating the correctness of each completion on a Juniper Ex3300 switch and verifying the syntax using configuration manuals. The dataset underwent several revisions to ensure consistent formatting, correct parameter extraction, and accurate outcomes. The dataset is designed to provide translations of high-level network intent to command-line configuration only. It does not target verifying the correctness of the parameters' values in the prompts or any other type of conflict detection or resolution. Despite this, all parameter values included in the dataset are valid values for the corresponding task. The dataset contains 1000 entries; each entry is presented using a JSON object of three elements: question, context, and answer. The question represents the high-level network intent, while the answer represents the corresponding low-level configurations. The context includes the generic form of the low-level configurations. The dataset is suitable for network intent translation research.

Keywords: Intent-based networks; Intent-translation; LLM; Network-configurations.

PubMed Disclaimer

Figures

Fig 1
Fig. 1
An example of a JSON object in the dataset.
Fig 2
Fig. 2
Subset of the hierarchy containing "traceoption" keyword.
Fig 3
Fig. 3
Completions format.
Fig 4
Fig. 4
Examples of prompt types.
Fig 5
Fig. 5
Commands distribution in the dataset.
See this image and copyright information in PMC

References

    1. Smarneh A. Hugging Face; 2024. NIT (Revision 7b200b0)
    1. Y. Miao, Y. Bai, L. Chen, D. Li, H. Sun, X. Wang, Z. Luo, Y. Ren, D. Sun and X. Xiuting, “An empirical study of netops capability of pre-trained large language models,” arXiv preprint arXiv:2309.05557, 2023.
    1. D. Donadel, F. Marchiori, L. Pajola and M. Conti, “Can LLMs understand computer networks? Towards a virtual system administrator,” arXiv preprint arXiv:2404.12689, 2024.
    1. Donadel D., Marchiori F., Pajola L., Conti M. GitHub; 2024. LLM-SysAdmin.
    1. B. Ifland, E. Duani, R. Krief, M. Ohana, A. Zilberman, A. Murillo, O. Manor, O. Lavi, H. Kenji and A. Shabtai, "Genet: a multimodal llm-based co-pilot for network topology and configuration," arXiv preprint arXiv:2407.08249, 2024.

LinkOut - more resources