Text2SQL is Not Enough: Unifying AI and Databases with TAG

Summary: Introduce Table‑Augmented Generation (TAG), a unified paradigm that generalizes Text2SQL and RAG to capture broader LM–DB interactions beyond relational‑algebra and point lookups. New TAG benchmarks show standard methods solve ≤20% of queries, evidencing the need for research on tightly integrating LM reasoning/knowledge with DBMS execution. (summarized by gpt-5-mini on Feb 09 2026)

Paper ID: 536
Venue: CIDR
Year: 2025
Pagerank: 8.0994951e-05
Overall Rank: 2,809 | 80.73%
DOI: 10.1145/nnnnnnn.nnnnnnn

Incoming Non-self Citations Over Time

Authors

1. Asim Biswal (University of California Berkeley)
2. Liana Patel (Stanford University)
3. Joseph E. Gonzalez (University of California Berkeley)
4. Siddharth Jha (University of California Berkeley)
5. Amog Kamsetty (University of California Berkeley)
6. Shu Liu (University of California Berkeley)
7. Carlos Guestrin (Stanford University)
8. Matei Zaharia (University of California Berkeley)

BibTeX Citation

@inproceedings{biswal_cidr25,
        address = {Amsterdam, Netherlands},
        series = {{CIDR} '25},
        title = {{Text2SQL is Not Enough: Unifying AI and Databases with TAG}},
        booktitle = {Proceedings of the {Conference} on {Innovative} {Data} {Systems} {Research}},
        author = {Biswal, Asim and Patel, Liana and Gonzalez, Joseph E. and Jha, Siddharth and Kamsetty, Amog and Liu, Shu and Guestrin, Carlos and Zaharia, Matei},
        year = {2025}
}

Incoming Citations (Sorted by Pagerank)

Showing 12 of 12 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
2,956	The Design of an LLM-powered Unstructured Analytics System	2025	CIDR	7.9203461e-05
4,045	Logical and Physical Optimizations for SQL Query Execution over Large Language Models	2025	SIGMOD	6.9394654e-05
5,485	Pneuma: Leveraging LLMs for Tabular Data Representation and Retrieval in an End-to-End System	2025	SIGMOD	6.2013279e-05
6,065	Vortex: Overcoming Memory Capacity Limitations in GPU-Accelerated Large-Scale Data Analytics	2025	VLDB	5.9895288e-05
7,568	Semantic Operators and Their Optimization: Enabling LLM-Based Data Processing with Accuracy Guarantees in LOTUS	2025	VLDB	5.5953379e-05
7,685	Beyond Quacking: Deep Integration of Language Models and RAG into DuckDB	2025	VLDB	5.5678545e-05
10,119	Data Movement-Aware GPU Sharing for Data-Intensive Systems	2026	CIDR	5.093636e-05
10,237	Factorized and Vectorized Execution: Optimizing Analytical and Semantic Queries over Relations	2026	SIGMOD	5.093636e-05
10,360	Drama: Unifying Data Retrieval and Analysis for Open-Domain Analytic Queries	2026	SIGMOD	5.093636e-05
10,405	AixelAsk: A Stepwise-Guided Retrieval and Reasoning Framework for Large Table QA	2026	SIGMOD	5.093636e-05
10,572	Relational Deep Dive: Error-Aware Queries Over Unstructured Data	2026	VLDB	5.093636e-05
10,618	ELT-Bench: An End-to-End Benchmark for Evaluating AI Agents on ELT Pipelines	2026	VLDB	5.093636e-05

Outgoing Citations (Sorted by Pagerank)

Showing 1 of 1 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
106	The MADlib Analytics Library or MAD Skills, the SQL	2012	VLDB	0.00033539462

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	2,602	NL2SQL is a solved problem... Not!	2024	CIDR
2	3,354	MT-Teql: Evaluating and Augmenting Neural NLIDB on Real-world Linguistic and Schema Variations	2022	VLDB
3	3,787	Combining Small Language Models and Large Language Models for Zero-Shot NL2SQL	2024	VLDB
4	5,174	An In-Depth Benchmarking of Text-to-SQL Systems	2021	SIGMOD
5	9,870	Natural Language to SQL: State of the Art and Open Problems	2025	VLDB
6	10,732	RTS+: Reliable Text to SQL	2025	SIGMOD
7	10,510	NL2SQLBench: A Modular Benchmarking Framework for LLM-Enabled NL2SQL Solutions	2026	VLDB
8	865	Natural language to SQL: Where are we today?	2020	VLDB
9	10,499	SchemaRAG: A Schema-aware Retrieval-Augmented Generation Framework for Text-to-SQL	2026	SIGMOD
10	279	Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation	2024	VLDB