Near-Duplicate Sequence Search at Scale for Large Language Model Memorization Evaluation

Summary: Proposes scalable near-duplicate sequence search to measure LLM memorization in trillion-token corpora. The approach groups min-hash values for all sequences with at least t tokens in linear time, uses inverted indexes and prefix filtering, and proves a bound 2^{(n+1)/(t+1)}−1, with real-world validation. (summarized by gpt-5-nano on Feb 09 2026)

Paper ID: 6744
Venue: SIGMOD
Year: 2023
Pagerank: 5.1757914e-05
Overall Rank: 10,024 | 31.23%
DOI: 10.1145/3589324

Incoming Non-self Citations Over Time

Authors

1. Zhencan Peng (Rutgers University)
2. Zhizhi Wang (Rutgers University)
3. Dong Deng (Rutgers University)

BibTeX Citation

@inproceedings{peng_sigmod23,
        title = {{Near-Duplicate Sequence Search at Scale for Large Language Model Memorization Evaluation}},
        author = {Peng, Zhencan and Wang, Zhizhi and Deng, Dong},
        series = {{SIGMOD} '23},
        booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
        publisher = {Association for Computing Machinery},
        doi = {10.1145/3589324},
        url = {https://dl.acm.org/doi/10.1145/3589324},
        year = {2023}
}

Incoming Citations (Sorted by Pagerank)

Showing 4 of 4 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
7,745	Near-Duplicate Text Alignment with One Permutation Hashing	2024	SIGMOD	5.5534781e-05
10,265	LSHAlign: All-Pair Near-Duplicate Text Alignment via LSH	2026	SIGMOD	5.093636e-05
10,554	Near-Duplicate Text Alignment under Weighted Jaccard Similarity	2026	VLDB	5.093636e-05
10,909	LLMLog: Advanced Log Template Generation via LLM-driven Multi-Round Annotation	2025	VLDB	5.093636e-05

Outgoing Citations (Sorted by Pagerank)

Showing 13 of 13 cited papers.

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

Rank	Cited Paper	Year	Venue	Pagerank
975	Can We Beat the Prefix Filtering? An Adaptive Framework for Similarity Join and Search	2012	SIGMOD	0.00012870645
978	Winnowing: Local Algorithms for Document Fingerprinting	2003	SIGMOD	0.00012859554
1,284	Copy Detection Mechanisms for Digital Documents	1995	SIGMOD	0.00011332822
1,596	Adaptive Parallel Aggregation Algorithms	1995	SIGMOD	0.00010247091
1,886	Pass-Join: A Partition-based Method for Similarity Joins	2012	VLDB	9.5358137e-05
3,474	An Efficient Partition Based Method for Exact Set Similarity Joins	2016	VLDB	7.3859271e-05
3,724	Overlap Set Similarity Joins with Theoretical Guarantees	2018	SIGMOD	7.1715735e-05
4,052	Local Similarity Search for Unstructured Text	2016	SIGMOD	6.935988e-05
5,412	Faerie: Efficient Filtering Algorithms for Approximate Dictionary-based Entity Extraction	2011	SIGMOD	6.2272563e-05
6,484	A Pivotal Prefix Based Filtering Algorithm for String Similarity Search	2014	SIGMOD	5.8665833e-05
7,676	Allign: Aligning All-Pair Near-Duplicate Passages in Long Texts	2021	SIGMOD	5.5686107e-05
8,522	TxtAlign: Efficient Near-Duplicate Text Alignment Search via Bottom-k Sketches for Plagiarism Detection	2022	SIGMOD	5.4119882e-05
9,705	META: An Efficient Matching-Based Method for Error-Tolerant Autocompletion	2016	VLDB	5.2351259e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	10,290	Skyline Retrieval meets Set-Cover Chunk Merging: A Cost-Effective RAG-Sketch for Long-Context LLM QA	2026	SIGMOD
2	10,265	LSHAlign: All-Pair Near-Duplicate Text Alignment via LSH	2026	SIGMOD
3	8,829	Cut Costs, Not Accuracy: LLM-Powered Data Processing with Guarantees	2026	SIGMOD
4	10,733	ScaleLLM: A Technique for Scalable LLM-augmented Data Systems	2025	SIGMOD
5	7,745	Near-Duplicate Text Alignment with One Permutation Hashing	2024	SIGMOD
6	13,343	Database Perspective on LLM Inference Systems	2025	VLDB
7	8,522	TxtAlign: Efficient Near-Duplicate Text Alignment Search via Bottom-k Sketches for Plagiarism Detection	2022	SIGMOD
8	10,318	In-context Clustering-based Entity Resolution with Large Language Models: A Design Space Exploration	2026	SIGMOD
9	9,310	LLM-PBE: Assessing Data Privacy in Large Language Models	2024	VLDB
10	8,908	Privacy and Accuracy-Aware AI/ML Model Deduplication	2025	SIGMOD