New paper proposes data-usage graphs to build trust in AI research datasets

A new paper proposes data-usage graphs as a method to enhance trust and find appropriate datasets for artificial intelligence research. These graphs map the connections between datasets and their usage in publications, software, and by researchers, revealing the "social life of data." The authors argue that this evidence, combined with production quality and provenance, can help determine a dataset's fitness for purpose. A prototype service within the National Data Platform demonstrates the feasibility of this approach. AI

IMPACT Proposes a new framework for improving data reliability in AI research, potentially enhancing the trustworthiness of AI-generated scientific results.

RANK_REASON The item is a research paper published on arXiv detailing a new methodology for data trust. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.IR (Information Retrieval) →

• alphaXiv
• artificial intelligence
• arXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• DagsHub
• data-usage graphs
• Gotit.pub
• Hugging Face
• Influence Flower
• Litmaps
• National Data Platform
• ScienceCast
• scite Smart Citations

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →