Predicate Pushdown for Data Science Pipelines

Summary: MagicPush uses a search-verification approach to predicate pushdown in data science pipelines, discovering input-space predicates and proving pushdown preserves outputs, even with non-relational operators and UDFs. Evaluations on TPC-H and 200 real-world GitHub Notebook pipelines show it beats a strong rule-based baseline, discovers new pushdown opportunities, and yields up to 99% running-time reduction in 42 pipelines while matching baseline opportunities elsewhere. (summarized by gpt-5-nano on Feb 09 2026)

Paper ID: 6701
Venue: SIGMOD
Year: 2023
Pagerank: 5.4194578e-05
Overall Rank: 8,465 | 41.93%
DOI: 10.1145/3589281

Incoming Non-self Citations Over Time

Authors

1. Cong Yan (Microsoft)
2. Yin Lin (University of Michigan)
3. Yeye He (Microsoft)

BibTeX Citation

@inproceedings{yan_sigmod23,
        title = {{Predicate Pushdown for Data Science Pipelines}},
        author = {Yan, Cong and Lin, Yin and He, Yeye},
        series = {{SIGMOD} '23},
        booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
        publisher = {Association for Computing Machinery},
        doi = {10.1145/3589281},
        url = {https://dl.acm.org/doi/10.1145/3589281},
        year = {2023}
}

Incoming Citations (Sorted by Pagerank)

Showing 4 of 4 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
9,917	The UDFBench Benchmark for General-purpose UDF Queries	2025	VLDB	5.1955087e-05
10,441	Data-Semantics-Aware Recommendation of Diverse Pivot Tables	2026	SIGMOD	5.093636e-05
10,690	Dynamic Pruning for Recursive Joins	2025	SIGMOD	5.093636e-05
11,078	LiquidCache: Efficient Pushdown Caching for Cloud-Native Data Analytics	2025	VLDB	5.093636e-05

Outgoing Citations (Sorted by Pagerank)

Showing 27 of 27 cited papers.

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

Rank	Cited Paper	Year	Venue	Pagerank
24	Spark SQL: Relational Data Processing in Spark	2015	SIGMOD	0.00054865648
132	Predicate Migration: Optimizing Queries with Expensive Predicates	1993	SIGMOD	0.00030378624
309	Optimization of Real Conjunctive Queries	1993	PODS	0.00021768596
894	Froid: Optimization of Imperative Programs in a Relational Database	2018	VLDB	0.00013367658
1,124	Query Optimization by Predicate Move-Around	1994	VLDB	0.00012087356
1,135	Qd-tree: Learning Data Layouts for Big Data Analytics	2020	SIGMOD	0.00012032847
1,691	PIVOT and UNPIVOT: Optimization and Execution Strategies in an RDBMS	2004	VLDB	9.9999141e-05
1,768	Tuplex: Data Science in Python at Native Code Speed	2021	SIGMOD	9.8041636e-05
1,889	SQL-on-Hadoop: Full Circle Back to Shared-Nothing Database Architectures	2014	VLDB	9.5335988e-05
1,982	WeTune: Automatic Discovery and Verification of Query Rewrite Rules	2022	SIGMOD	9.3573897e-05
2,202	Quantifying TPC-H Choke Points and Their Optimizations	2020	VLDB	8.9639459e-05
2,391	Mison: A Fast JSON Parser for Data Analytics	2017	VLDB	8.6413407e-05
2,563	AnalyticDB: Real-time OLAP Database System at Alibaba Cloud	2019	VLDB	8.412445e-05
2,651	Magpie: Python at Speed and Scale using Cloud Backends	2021	CIDR	8.2918086e-05
2,744	Auto-Suggest: Learning-to-Recommend Data Preparation Steps Using Data Science Notebooks	2020	SIGMOD	8.1781662e-05
3,137	Pushing Data-Induced Predicates Through Joins in Big-Data Clusters	2020	VLDB	7.7204167e-05
3,172	Demonstration of the Cosette Automated SQL Prover	2017	SIGMOD	7.6679093e-05
3,327	Automated Verification of Query Equivalence Using Satisfiability Modulo Theories	2019	VLDB	7.518491e-05
3,632	FlexPushdownDB: Hybrid Pushdown and Caching in a Cloud DBMS	2021	VLDB	7.2368817e-05
3,992	Aggify: Lifting the Curse of Cursor Loops using Custom Aggregates	2020	SIGMOD	6.9695338e-05
4,599	Automatically Leveraging MapReduce Frameworks for Data-Intensive Applications	2018	SIGMOD	6.6122875e-05
5,348	Crystal: A Unified Cache Storage System for Analytical Databases	2021	VLDB	6.2562688e-05
6,031	Sia: Optimizing Queries using Learned Predicates	2021	SIGMOD	6.0007422e-05
6,194	Incorporating Super-Operators in Big-Data Query Optimizers	2020	VLDB	5.9470844e-05
6,314	YeSQL: "You extend SQL" with Rich and Highly Performant User-Defined Functions in Relational Databases	2022	VLDB	5.9157364e-05
7,035	Optimizing Recursive Queries with Program Synthesis	2022	SIGMOD	5.7216112e-05
9,966	Generating Application-Specific Data Layouts for In-memory Databases	2019	VLDB	5.1870939e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	7,176	Interactive Demonstration of Probabilistic Predicates	2018	SIGMOD
2	2,865	End-to-end Optimization of Machine Learning Prediction Queries	2022	SIGMOD
3	7,139	Selection Pushdown in Column Stores using Bit Manipulation Instructions	2023	SIGMOD
4	4,114	Optimizing Machine Learning Inference Queries with Correlative Proxy Models	2022	VLDB
5	4,575	Conjunctive Queries with Comparisons	2022	SIGMOD
6	3,137	Pushing Data-Induced Predicates Through Joins in Big-Data Clusters	2020	VLDB
7	1,124	Query Optimization by Predicate Move-Around	1994	VLDB
8	132	Predicate Migration: Optimizing Queries with Expensive Predicates	1993	SIGMOD
9	11,166	PLAQUE: Automated Predicate Learning at Query Time	2024	SIGMOD
10	295	Accelerating Machine Learning Inference with Probabilistic Predicates	2018	SIGMOD