Tracing the Data Trail: A Survey of Data Provenance, Transparency and Traceability in LLMs

TL;DR

This survey reviews a decade of research on data provenance, transparency, and traceability in large language models, highlighting methodologies, challenges, and a new taxonomy for the field.

Contribution

It introduces a comprehensive taxonomy of data provenance and transparency in LLMs, synthesizing 95 publications and identifying key methodologies and trade-offs.

Findings

Key methodologies include data watermarking and bias measurement

Trade-offs exist between transparency and data privacy

A taxonomy for data provenance in LLMs is proposed

Abstract

Large language models (LLMs) are deployed at scale, yet their training data life cycle remains opaque. This survey synthesizes research from the past ten years on three tightly coupled axes: (1) data provenance, (2) transparency, and (3) traceability, and three supporting pillars: (4) bias \& uncertainty, (5) data privacy, and (6) tools and techniques that operationalize them. A central contribution is a proposed taxonomy defining the field's domains and listing corresponding artifacts. Through analysis of 95 publications, this work identifies key methodologies concerning data generation, watermarking, bias measurement, data curation, data privacy, and the inherent trade-off between transparency and opacity.