Multi-Objective Agentic Rewrites for Unstructured Data Processing
Summary: MOAR optimizes LLM-powered DocETL pipelines via 30+ cost/accuracy rewrite directives and global, multi-armed-bandit search over interacting rewrites. Across six workloads, it delivers 27% higher accuracy than ABACUS, or comparable accuracy at 55% of the cost. (summarized by gpt-5.6-luna on Aug 28 2026)
Incoming Non-self Citations Over Time
Authors
- 1. Lindsey Linxi Wei (University of Washington)
- 2. Shreya Shankar (University of California Berkeley)
- 3. Sepanta Zeighami (University of California Berkeley)
- 4. Yeounoh Chung (Google)
- 5. Fatma Ozcan (Google)
- 6. Aditya G. Parameswaran (University of California Berkeley)
BibTeX Citation
@article{wei_vldb26,
title = {{Multi-Objective Agentic Rewrites for Unstructured Data Processing}},
author = {Wei, Lindsey Linxi and Shankar, Shreya and Zeighami, Sepanta and Chung, Yeounoh and Ozcan, Fatma and Parameswaran, Aditya G.},
journal = {PVLDB},
series = {{VLDB} '26},
volume = {19},
number = {11},
pages = {3062--3075},
doi = {10.14778/3836663.3836673},
url = {https://doi.org/10.14778/3836663.3836673},
year = {2026}
}Incoming Citations (Sorted by Pagerank)
Showing 2 of 2 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 10,346 | KathDB: Explainable Multimodal Database Management System with Human-AI Collaboration | 2026 | CIDR | 4.9793485e-05 |
| 10,980 | Carnot: Interpretable, Interactive, and Optimized Execution of Deep Research Queries | 2026 | VLDB | 4.9793485e-05 |
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 22 of 22 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
Previous
Page 1 / 1
Next
Semantically Similar Papers
| # | Overall Rank | Paper | Year | Venue |
|---|---|---|---|---|
| 1 | 6,869 | Doctopus: Budget-aware Structural Table Extraction from Unstructured Documents | 2025 | VLDB |
| 2 | 10,353 | Making Prompts First-Class Citizens for Adaptive LLM Pipelines | 2026 | CIDR |
| 3 | 9,066 | MoDora: Tree-Based Semi-Structured Document Analysis System | 2026 | SIGMOD |
| 4 | 10,832 | ReSequel: Robust LLM-assisted Query Rewriting and Optimization using Templatization and Sampling | 2026 | VLDB |
| 5 | 3,695 | Revisiting Prompt Engineering via Declarative Crowdsourcing | 2024 | CIDR |
| 6 | 8,997 | Cut Costs, Not Accuracy: LLM-Powered Data Processing with Guarantees | 2026 | SIGMOD |
| 7 | 11,061 | LLM-AutoDP: Automatic Data Processing via LLM Agents for Model Fine-tuning | 2026 | VLDB |
| 8 | 3,126 | Abacus: A Cost-Based Optimizer for Semantic Operator Systems | 2026 | VLDB |
| 9 | 10,690 | Task Cascades for Efficient Unstructured Data Processing | 2026 | SIGMOD |
| 10 | 683 | DocETL: Agentic Query Rewriting and Evaluation for Complex Document Processing | 2025 | VLDB |