Reproduce “SABER: A SQL-Compatible Semantic Document Processing System Based on Extended Relational Algebra” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Changjae Lee, Zhuoyue Zhao, Jinjun Xiong
Researchers at the University at Buffalo developed SABER, a SQL-compatible semantic document processing system that extends relational algebra with 10 LLM-backed semantic operators. This framework enables principled, optimizable querying across mixed structured and unstructured data, addressing the lack of a formal foundation in existing semantic data processing systems.
Key insights
- Semantic operators can be formally defined to retain the algebraic properties of their relational counterparts, enabling the application of existing query optimization rules.
- A SQL-compatible, UDF-based approach allows declarative integration of LLM capabilities into traditional database queries without extensive parser modifications.
- Existing SDPSs, despite their individual limitations, can be modularly leveraged as physical backends within a unified logical semantic algebra, promoting interoperability.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.