DemandClean: A Multi-Objective Learning Framework for Balancing Model Tolerance to Data Authenticity and Diversity
Summary: DemandClean: an RL framework that adaptively picks Repair/Delete/No to trade off data authenticity, feature diversity, and downstream models' noise tolerance. By leveraging error types (missing/semantic/syntactic) and interpretable visualizations, it matches or improves accuracy while cutting repair/deletion actions ≈80% vs Repair‑All, greatly reducing preprocessing cost. (summarized by gpt-5-mini on Feb 09 2026)
Incoming Non-self Citations Over Time
No non-self incoming citations found for this paper in this database.
Authors
- 1. Zekai Qian (Harbin Engineering University)
- 2. Xiaoou Ding (Harbin Engineering University)
- 3. Chen Wang (Tsinghua University)
- 4. Hongzhi Wang (Harbin Engineering University)
BibTeX Citation
@article{qian_vldb25,
title = {{DemandClean: A Multi-Objective Learning Framework for Balancing Model Tolerance to Data Authenticity and Diversity}},
author = {Qian, Zekai and Ding, Xiaoou and Wang, Chen and Wang, Hongzhi},
journal = {PVLDB},
series = {{VLDB} '25},
volume = {18},
number = {12},
pages = {5339--5342},
doi = {10.14778/3750601.3750666},
url = {https://doi.org/10.14778/3750601.3750666},
year = {2025}
}Incoming Citations (Sorted by Pagerank)
Showing 0 of 0 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 5 of 5 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
| Rank | Cited Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 112 | HoloClean: Holistic Data Repairs with Probabilistic Inference | 2017 | VLDB | 0.00032801121 |
| 3,580 | Automatic Data Repair: Are We Ready to Deploy? | 2024 | VLDB | 7.2888516e-05 |
| 3,886 | GoodCore: Data-effective and Data-efficient Machine Learning through Coreset Selection over Incomplete Data | 2023 | SIGMOD | 7.0460597e-05 |
| 9,697 | Clean4TSDB: A Data Cleaning Tool for Time Series Databases | 2024 | VLDB | 5.2351259e-05 |
| 9,699 | MTSClean: Efficient Constraint-based Cleaning for Multi-Dimensional Time Series Data | 2024 | VLDB | 5.2351259e-05 |
Previous
Page 1 / 1
Next
Semantically Similar Papers
| # | Overall Rank | Paper | Year | Venue |
|---|---|---|---|---|
| 1 | 2,981 | Towards Dependable Data Repairing with Fixing Rules | 2014 | SIGMOD |
| 2 | 2,041 | Combining Quantitative and Logical Data Cleaning | 2016 | VLDB |
| 3 | 11,746 | Active Reinforcement Learning for Data Preparation: Learn2Clean with Human-In-The-Loop | 2020 | CIDR |
| 4 | 112 | HoloClean: Holistic Data Repairs with Probabilistic Inference | 2017 | VLDB |
| 5 | 3,580 | Automatic Data Repair: Are We Ready to Deploy? | 2024 | VLDB |
| 6 | 7,880 | Learning Over Dirty Data Without Cleaning | 2020 | SIGMOD |
| 7 | 1,323 | Data Cleaning: Overview and Emerging Challenges | 2016 | SIGMOD |
| 8 | 10,785 | Auto-Test: Learning Semantic-Domain Constraints for Unsupervised Error Detection in Tables | 2025 | SIGMOD |
| 9 | 9,590 | Constraint-Variance Tolerant Data Repairing | 2016 | SIGMOD |
| 10 | 5,794 | ActiveClean: An Interactive Data Cleaning Framework For Modern Machine Learning | 2016 | SIGMOD |