Task Cascades for Efficient Unstructured Data Processing
Summary: Task cascades generalize model cascades for LLM-based document processing by varying not only the model, but also the queried span and even the operation, exploiting simpler correlated sub-tasks and partial evidence. An iterative optimizer plus statistical accuracy guarantees yields 36% lower cost than standard cascades at 90% target accuracy. (summarized by gpt-5.4-mini on Apr 11 2026)
Incoming Non-self Citations Over Time
Authors
- 1. Shreya Shankar (University of California Berkeley)
- 2. Sepanta Zeighami (University of California Berkeley)
- 3. Aditya G. Parameswaran (University of California Berkeley)
BibTeX Citation
@inproceedings{shankar_sigmod26,
title = {{Task Cascades for Efficient Unstructured Data Processing}},
author = {Shankar, Shreya and Zeighami, Sepanta and Parameswaran, Aditya G.},
series = {{SIGMOD} '26},
booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
publisher = {Association for Computing Machinery},
doi = {10.1145/3786702},
url = {https://dl.acm.org/doi/10.1145/3786702},
year = {2026}
}
Incoming Citations (Sorted by Pagerank)
Showing 2 of 2 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 6,607 | Multi-Objective Agentic Rewrites for Unstructured Data Processing | 2026 | VLDB | 5.73539e-05 |
| 10,351 | Evergreen: Efficient Claim Verification for Semantic Aggregates | 2027 | VLDB | 4.9769913e-05 |
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 29 of 29 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 | 496 | Language Models Enable Simple Systems for Generating Structured Views of Heterogeneous Data Lakes | 2024 | VLDB |
| 2 | 8,897 | Optimized Batch Prompting for Cost-effective LLMs | 2025 | VLDB |
| 3 | 8,688 | Unstructured Data Analysis Using LLMs: A Comprehensive Benchmark | 2026 | VLDB |
| 4 | 11,535 | Unstructured Data Fusion for Schema and Data Extraction | 2024 | SIGMOD |
| 5 | 11,170 | ScaleLLM: A Technique for Scalable LLM-augmented Data Systems | 2025 | SIGMOD |
| 6 | 8,182 | DocDB: A Database for Unstructured Document Analysis | 2025 | VLDB |
| 7 | 658 | DocETL: Agentic Query Rewriting and Evaluation for Complex Document Processing | 2025 | VLDB |
| 8 | 4,565 | QUEST: Query Optimization in Unstructured Document Analysis | 2025 | VLDB |
| 9 | 10,509 | ScaleDoc: Scaling LLM-based Predicates over Large Document Collections | 2026 | SIGMOD |
| 10 | 8,526 | Cut Costs, Not Accuracy: LLM-Powered Data Processing with Guarantees | 2026 | SIGMOD |