Database Paper Browser

Back to papers

AOP: Automated and Interactive LLM Pipeline Orchestration for Answering Complex Queries

Summary: AOP: first system for automated, interactive LLM pipeline orchestration on data lakes; defines reusable semantic operators (retrieval, filter, aggregate, validate) to assemble multi‑step workflows across unstructured, semi‑structured, and structured data. LLM‑driven operator extraction plus pipeline optimizations (prefetching, parallelism, interactive self‑reflection) yields large accuracy gains (≈45%) on complex query benchmarks. (summarized by gpt-5-mini on Feb 09 2026)

Paper ID
560
Venue
CIDR
Year
2025
Pagerank
4.6730494e-05
Overall Rank
7,705 | 46.40%
DOI
-

Incoming Non-self Citations Over Time

Authors

Incoming Citations (Sorted by Pagerank)

Showing 7 of 7 citing papers.

Previous Page 1 / 1 Next

Outgoing Citations (Sorted by Pagerank)

Showing 7 of 7 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