Prism: Private Relational Data Synthesis with Language Models

Summary: Prism synthesizes relational data under differential privacy by distilling an ensemble of teacher LLMs into a student generator. Trie-based token pruning, adaptive noise, and thresholded sampling reduce privacy cost while preserving utility and diversity. (summarized by gpt-5.6-luna on Jul 26 2026)

Paper ID: 7475
Venue: SIGMOD
Year: 2026
Pagerank: 5.093636e-05
Overall Rank: 10,282 | 29.46%
DOI: 10.1145/3802102

Incoming Non-self Citations Over Time

No non-self incoming citations found for this paper in this database.

Authors

1. Guohui Guan (University of Minnesota)
2. Chang Ge (University of Minnesota)

BibTeX Citation

@inproceedings{guan_sigmod26,
        title = {{Prism: Private Relational Data Synthesis with Language Models}},
        author = {Guan, Guohui and Ge, Chang},
        series = {{SIGMOD} '26},
        booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
        publisher = {Association for Computing Machinery},
        doi = {10.1145/3802102},
        url = {https://dl.acm.org/doi/10.1145/3802102},
        year = {2026}
}

Incoming Citations (Sorted by Pagerank)

Showing 0 of 0 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank

Outgoing Citations (Sorted by Pagerank)

Showing 14 of 14 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
89	On the Propagation of Errors in the Size of Join Results	1991	SIGMOD	0.00035031529
121	Boosting the Accuracy of Differentially Private Histograms Through Consistency	2010	VLDB	0.00031639377
281	Towards Practical Differential Privacy for SQL Queries	2018	VLDB	0.00022445849
376	Discovering Denial Constraints	2013	VLDB	0.00019677674
1,144	PrivateSQL: A Differentially Private SQL Query Engine	2019	VLDB	0.00011999046
1,500	Efficient Discovery of Approximate Dependencies	2018	VLDB	0.00010561098
2,476	Data Synthesis via Differentially Private Markov Random Fields	2021	VLDB	8.5258582e-05
2,841	AIM: An Adaptive and Iterative Mechanism for Differentially Private Synthetic Data	2022	VLDB	8.0637668e-05
3,412	Kamino: Constraint-Aware Differentially Private Data Synthesis	2021	VLDB	7.4344343e-05
4,881	Utility Cost of Formal Privacy for Releasing National Employer-Employee Statistics	2017	SIGMOD	6.465615e-05
5,055	APEx: Accuracy-Aware Differentially Private Data Exploration	2019	SIGMOD	6.3824509e-05
6,555	Controllable Tabular Data Synthesis Using Diffusion Models	2024	SIGMOD	5.8415111e-05
7,338	DPGraph: A Benchmark Platform for Differentially Private Graph Analysis	2021	SIGMOD	5.6393827e-05
8,719	DPXPlain: Privately Explaining Aggregate Query Answers	2023	VLDB	5.3774243e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	3,016	Plausible Deniability for Privacy-Preserving Data Synthesis	2017	VLDB
2	2,476	Data Synthesis via Differentially Private Markov Random Fields	2021	VLDB
3	4,827	PrivLava: Synthesizing Relational Data with Foreign Keys under Differential Privacy	2023	SIGMOD
4	10,335	A General Framework for Per-record Differential Privacy	2026	SIGMOD
5	4,261	Relational Data Synthesis using Generative Adversarial Networks: A Design Space Exploration	2020	VLDB
6	10,345	Benchmarking Differentially Private Tabular Data Synthesis: [Experiments & Analysis]	2026	SIGMOD
7	3,412	Kamino: Constraint-Aware Differentially Private Data Synthesis	2021	VLDB
8	10,775	PrivPetal: Relational Data Synthesis via Permutation Relations	2025	SIGMOD
9	10,483	PRISM: Navigating Cost–Accuracy Trade-offs for NL2SQL	2026	SIGMOD
10	9,310	LLM-PBE: Assessing Data Privacy in Large Language Models	2024	VLDB