Open Datasheets: Machine-readable Documentation for Open Datasets and Responsible AI Assessments

arXiv:2312.06153v2 Announce Type: replace
Abstract: This paper introduces a no-code, machine-readable documentation framework for open datasets, with a focus on responsible AI (RAI) considerations. The framework aims to improve comprehensibility, and usability of open datasets, facilitating easier discovery and use, better understanding of content and context, and evaluation of dataset quality and accuracy. The proposed framework is designed to streamline the evaluation of datasets, helping researchers, data scientists, and other open data users quickly identify datasets that meet their needs …

abstract arxiv code context cs.ai cs.hc cs.lg datasets discovery documentation evaluation focus framework machine no-code paper responsible responsible ai type understanding usability