Skip to content
Home

Selected Earlier Work · Self-Improving Data Systems

Database Learning: QuickSel and BlinkML

Database-learning methods that improve estimation and reduce computation using workload history.

QuickSel

BlinkML

Research question

What if a database could continuously learn from the workload it already sees?

Approach

Use workload-driven learning, compact probabilistic models, and statistically bounded computation to improve future decisions.

Key results

  • Database Learning reuses knowledge from prior query answers to accelerate future queries.
  • QuickSel learns selectivity distributions from query feedback using mixture models.
  • BlinkML accelerates maximum-likelihood training by 6.26×–629× while guaranteeing, with 95% probability, the same predictions as full-data training.

Publications

Database Learning: Toward a Database that Becomes Smarter Every Time

Yongjoo Park, Ahmad Shahab Tajik, Michael J. Cafarella, Barzan Mozafari

SIGMOD 2017 · ACM SIGMOD International Conference on Management of Data

Paper ↗Project ↗
Cite
@inproceedings{3cebf7e2-930f-4c8a-aa17-75a4c32cb001,
  title = {Database Learning: Toward a Database that Becomes Smarter Every Time},
  author = {Yongjoo Park and Ahmad Shahab Tajik and Michael J. Cafarella and Barzan Mozafari},
  booktitle = {ACM SIGMOD International Conference on Management of Data},
  year = {2017},
  doi = {10.1145/3035918.3064013}
}

BlinkML: Efficient Maximum Likelihood Estimation with Probabilistic Guarantees

Yongjoo Park, Jingyi Qing, Xiaoyang Shen, Barzan Mozafari

SIGMOD 2019 · ACM SIGMOD International Conference on Management of Data

Paper ↗Technical Report ↗Project ↗
Cite
@inproceedings{50db23fc-a5a3-4d7f-ac07-2a696129c92f,
  title = {BlinkML: Efficient Maximum Likelihood Estimation with Probabilistic Guarantees},
  author = {Yongjoo Park and Jingyi Qing and Xiaoyang Shen and Barzan Mozafari},
  booktitle = {ACM SIGMOD International Conference on Management of Data},
  year = {2019},
  doi = {10.1145/3299869.3300077}
}

QuickSel: Quick Selectivity Learning with Mixture Models

Yongjoo Park, Shucheng Zhong, Barzan Mozafari

SIGMOD 2020 · ACM SIGMOD International Conference on Management of Data

Paper ↗Technical Report ↗Project ↗
Cite
@inproceedings{ea3d8506-1f2f-40ea-8bcb-6a74f72e12f0,
  title = {QuickSel: Quick Selectivity Learning with Mixture Models},
  author = {Yongjoo Park and Shucheng Zhong and Barzan Mozafari},
  booktitle = {ACM SIGMOD International Conference on Management of Data},
  year = {2020},
  doi = {10.1145/3318464.3389727}
}