Fine-Grained Table Retrieval Through the Lens of Complex Queries

Wojciech Kosiuk; Xingyu Ji; Yeounoh Chung; Fatma Özcan; Madelon Hulsebos

cs.IRMar 7, 2026

Fine-Grained Table Retrieval Through the Lens of Complex Queries

Authors: Wojciech Kosiuk, Xingyu Ji, Yeounoh Chung, Fatma Özcan, Madelon Hulsebos

Organizations: University of Amsterdam · UC Berkeley · Google · CWI

Abstract

Enabling question answering over tables and databases in natural language has become a key capability in the democratization of insights from tabular data sources. These systems first require retrieval of data that is relevant to a given natural language query, for which several methods have been introduced. In this work we present and study a table retrieval mechanism devising fine-grained typed query decomposition and global connectivity-awareness (DCTR), to handle the challenges induced by open-domain question answering over relational databases in complex usage contexts. We evaluate the effectiveness of the two mechanisms through the lens of retrieval complexity which we measure along the axes of query- and data complexity. Our analyses over industry-aligned benchmarks illustrate the robustness of DCTR for highly composite queries and densely connected databases.

Figures & tables

Figure 1: Overview of DCTR with component-wise retrieval, group formation, and FK-group expansion.

Setting	Model	BEAVER	FIBEN	BIRD
Baseline	stella	0.181	0.297	0.901
	bge-small	0.154	0.198	0.809
	e5-small	0.108	0.062	0.737
$\text{vote}_{k}\!=\!1$ , $n_{\text{groups}}\!=\!1$	stella	0.231 $\pm$ .004	0.317 $\pm$ .001	0.834 $\pm$ .001
	bge-small	0.223 $\pm$ .005	0.294 $\pm$ .003	0.826 $\pm$ .002
	e5-small	0.199 $\pm$ .009	0.257 $\pm$ .004	0.817 $\pm$ .007

Table 1: Capped Recall@5 (mean

\pm

std) across datasets and models. Bold values indicate the best result per dataset and embedding model.

Setting	Model	BEAVER	FIBEN	BIRD
Baseline	stella	0.277	0.413	0.961
	bge-small	0.231	0.262	0.909
	e5-small	0.175	0.140	0.850
$\text{vote}_{k}\!=\!2$ , $n_{\text{groups}}\!=\!2$	stella	0.313 $\pm$ .007	0.396 $\pm$ .006	0.899 $\pm$ .004
	bge-small	0.326 $\pm$ .002	0.331 $\pm$ .006	0.893 $\pm$ .003
	e5-small	0.273 $\pm$ .007	0.346 $\pm$ .004	0.882 $\pm$ .006

Table 2: Capped Recall@10 (mean

\pm

std) across datasets and models. Bold values indicate the best result per dataset and embedding model.

Setting	Model	BEAVER	FIBEN	BIRD
Baseline	stella	0.413	0.584	0.986
	bge-small	0.350	0.461	0.976
	e5-small	0.281	0.283	0.949
$\text{vote}_{k}\!=\!5$ , $n_{\text{groups}}\!=\!5$	stella	0.435 $\pm$ .002	0.513 $\pm$ .007	0.971 $\pm$ .002
	bge-small	0.429 $\pm$ .003	0.448 $\pm$ .001	0.972 $\pm$ .002
	e5-small	0.408 $\pm$ .008	0.516 $\pm$ .003	0.965 $\pm$ .001

Table 3: Capped Recall@25 (mean

\pm

std) across datasets and models. Bold values indicate the best result per dataset and embedding model.

Figure 2: Query length vs recall@25, aggregated across datasets. Comparison through three embedding models, shows that DCTR outperforms the baseline, especially for longer queries (

\geq

40 tokens).

Figure 3: Number of components vs Recall@25 across datasets. Comparison through three embedding models shows that DCTR is robust for compositional queries while a single-vector baseline does not resolve compositional queries regardless of model capacity.

Figure 4: Number of tables connected to gold tables vs recall@25, combined across datasets, for different embedding models. DCTR improves more on queries where gold tables are densely connected.

Appendix figures & tables6 assets

Supplementary material from the paper’s appendix.

Appendix

Panel of Figure 5 — Figure 5: Query length vs recall@k, combined across datasets. Comparison of three embedding models for the proposed method versus the baseline. Across all values of $k$ , the proposed method consistently outperforms the baseline, especially for longer queries ( $\geq$ 40 tokens).

Panel of Figure 6 — Figure 6: Number of components vs recall@k, combined across datasets. Comparison of three embedding models for the proposed method versus the baseline. Our method consistently performs better than the baseline for queries with more components. The proposed method consistently outperforms the baseline as the number of query components increases. At $k{=}25$ (Figure 6(c) ), our method maintains stable performance for highly compositional queries, whereas the baseline exhibits a performance drop.

Panel of Figure 7 — Figure 7: Number of tables gold tables vs recall@k, combined across datasets. Comparison of three embedding models for the proposed method versus the baseline. Both methods exhibit a degradation in capped recall as there the number of gold tables increases. The proposed method behaves similarly to the baseline, indicating limited advantage when there is a lot of relevant tables to retrieve.

Panel of Figure 8 — Figure 8: Number of gold tables join relationships vs recall@k, combined across datasets. Comparison of three embedding models for the proposed method versus the baseline. DCTR improves performance on queries for which gold tables are densely connected.

Panel of Figure 9 — Figure 9: Distribution of analyzed properties. Results combined for all datasets.

Metric	BEAVER	BIRD	FIBEN
Queries	209	1534	300
Unique DBs	6	11	1
Avg query (w)	31.9	14.6	12.3
Avg SQL (tok)	152.3	25.4	61.9
Tables	463	75	152
Columns	4238	798	373

Appendix

Table 7: Benchmark datasets characteristics.

Table Retrieval

Table retrieval finds relevant tables and schema elements for question answering, analytics, and text-to-SQL.

See all 19 papers

JoinGR: Learning to Traverse Join Graphs for Table Retrieval
Oct 1, 2026Sandipan De, Abhijit Chakraborty, Sambaran Bandyopadhyay +1Graph-Based Retrieval Table Retrieval
Arizona State University · Adobe Research, India
FGTR: Fine-Grained Multi-Table Retrieval via Hierarchical LLM Reasoning
Mar 13, 2026Chaojie Sun, Bin Cao, Tiantian Li +3Table Retrieval LLM Reasoning
Zhejiang University of Technology Hangzhou, China · University of Aberdeen Aberdeen, UK
Retrieve Only Relevant Tables Whether Few or Many: Adaptive Table Retrieval Method
Apr 12, 2026Taehee Kim, Seungbin Yang, Jihwan Kim +1Table Retrieval Adaptive Retrieval
KAIST AI · Letsur

Oct 1, 2026J/K move · Enter open · S save

JoinGR: Learning to Traverse Join Graphs for Table Retrieval

Sandipan De, Abhijit Chakraborty, Sambaran Bandyopadhyay, Vivek Gupta

Arizona State University · Adobe Research, India

Graph-Based Retrieval Table Retrieval Database Systems

Open paper AskPDF arXiv

Retrieving the right tables is a prerequisite for Text-to-SQL over realistic databases. Dense table retrievers rank schema elements independently, but this ignores a key source of evidence: some required tables are not mentioned in the question and become identifiable only through their join relationships to already relevant tables. We introduce JOINGR, a join-aware table retrieval method that treats the database join graph as the retrieval space. Columns are represented as graph nodes, while intra-table and foreign-key relationships are represented as typed edges. Given a question, JOINGR selects semantically similar anchor tables, traverses join edges with a query-conditioned scorer, and aggregates the resulting edge deposits into table scores. The scorer is a lightweight MLP on top of frozen query, node, and edge embeddings, trained with a pairwise margin loss over gold tables. On BIRD and Spider datasets, JOINGR is competitive with the strongest retrieval baselines. On BEAVER, a challenging enterprise benchmark with multi-hop table requirements, JOINGR substantially improves recall over dense retrieval and re-ranking baselines. Cross-domain experiments show that the learned scorer transfers across benchmarks, indicating that the method captures reusable joingraph traversal behavior.

Abstract

Table Retrieval

Explore similar work

JoinGR: Learning to Traverse Join Graphs for Table Retrieval

FGTR: Fine-Grained Multi-Table Retrieval via Hierarchical LLM Reasoning

Retrieve Only Relevant Tables Whether Few or Many: Adaptive Table Retrieval Method