SynthTextEval: Synthetic Text Data Generation and Evaluation for High-Stakes Domains

2025/07/09 by Krithika Ramesh, Ramesh, Krithika, Daniel Smolyak +12 · 1 voice · 1 citation
Computer Science · Decision Sciences · Medicine · Social Sciences · #Artificial Intelligence in Healthcare and Education #Computation and Language (cs.CL) #Computational and Text Analysis Methods #FOS: Computer and information sciences #Scientific Computing and Data Management #cs.CL

paper · pdf · doi:10.48550/arxiv.2507.07229

openalex publication_date 2025/07/09 · arxiv published 2025/07/09 · openalex created_date 2025/10/10 · arxiv updated 2025/11/03 · openalex updated_date 2026/07/28

Abstract

We present SynthTextEval, a toolkit for conducting comprehensive evaluations of synthetic text. The fluency of large language model (LLM) outputs has made synthetic text potentially viable for numerous applications, such as reducing the risks of privacy violations in the development and deployment of AI systems in high-stakes domains. Realizing this potential, however, requires principled consistent evaluations of synthetic data across multiple dimensions: its utility in downstream systems, the fairness of these systems, the risk of privacy leakage, general distributional differences from the source text, and qualitative feedback from domain experts. SynthTextEval allows users to conduct evaluations along all of these dimensions over synthetic data that they upload or generate using the toolkit's generation module. While our toolkit can be run over any data, we highlight its functionality and effectiveness over datasets from two high-stakes domains: healthcare and law. By consolidating and standardizing evaluation metrics, we aim to improve the viability of synthetic text, and in-turn, privacy-preservation in AI development.

Citations

• A False Sense of Privacy: Evaluating Textual Data Sanitization Beyond Surface-level Privacy Leakage
• SafeSynthDP: Leveraging Large Language Models for Privacy-Preserving Synthetic Data Generation Using Differential Privacy
• Evaluating Differentially Private Synthetic Data Generation in High-Stakes Domains
• Evaluating Synthetic Data Generation from User Generated Text
• Privacy Evaluation Benchmarks for NLP Models
• On LLMs-Driven Synthetic Data Generation, Curation, and Evaluation: A Survey
• Rethinking LLM Memorization through the Lens of Adversarial Compression
• Large Language Models for Data Annotation and Synthesis: A Survey
• DataDreamer: A Tool for Synthetic Data Generation and Reproducible LLM Workflows
• Orca 2: Teaching Small Language Models How to Reason
• PrivLM-Bench: A Multi-level Privacy Evaluation Benchmark for Language Models
• Assessment of Differentially Private Synthetic Data for Utility and Fairness in End-to-End Machine Learning Pipelines for Tabular Data
• Privacy-Preserving In-Context Learning with Differentially Private Few-Shot Generation
• FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generation
• Auditing and Generating Synthetic Data with Controllable Trust Trade-offs
• Synthcity: facilitating innovative use cases of synthetic data in different data modalities
• On the Blind Spots of Model-Based Evaluation Metrics for Text Generation
• Synthetic Text Generation with Differential Privacy: A Simple and Practical Recipe
• Differentially Private Language Models for Secure Data Sharing
• Large Language Models Can Self-Improve
• Mention Annotations Alone Enable Efficient Domain Adaptation for Coreference Resolution
• The Text Anonymization Benchmark (TAB): A Dedicated Corpus and Evaluation Framework for Text Anonymization
• The Text Anonymization Benchmark (TAB): A Dedicated Corpus and Evaluation Framework for Text Anonymization
• Numerical Composition of Differential Privacy
• MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontiers
• The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics
• CTRL: A Conditional Transformer Language Model for Controllable Generation
• The Curious Case of Neural Text Degeneration
• Ensuring Fairness in Machine Learning to Advance Health Equity
• Hierarchical Neural Story Generation
• The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networks

Cited by

• Controlled Generation for Private Synthetic Text

Discussions

• Thank you to @anjalief.bsky.social for advising. Hands-on with DP-SGD? Start with our another paper and open-source package (arxiv.org/abs/2507.07229 github.com/kr-ramesh/sy...) [bsky, 3 points, 0 comments]

Related

• Evaluating Synthetic Data Generation from User Generated Text
• A Multi-Faceted Evaluation Framework for Assessing Synthetic Data Generated by Large Language Models
• SynBench: A Benchmark for Differentially Private Text Generation
• Evaluating Differentially Private Synthetic Data Generation in High-Stakes Domains
• Best Practices and Lessons Learned on Synthetic Data