S2SQL: Injecting Syntax to Question-Schema Interaction Graph Encoder for Text-to-SQL Parsers

Hui, Binyuan; Geng, Ruiying; Wang, Lihan; Qin, Bowen; Li, Yanyang; Li, Bowen; Sun, Jian; Li, Yongbin

doi:10.18653/v1/2022.findings-acl.99

Cited by 29 publications

(23 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Results are reported in Table 3. We can observe that in all three datasets, RESDSQL-3B + NatSQL surprisingly outperforms all strong competitors by a large margin, which suggests that our decoupling idea can also improve (Gan et al 2021b) 73.7 75.0 68.7 73.3 SMBOP + GRAPPA (Rubin and Berant 2021) 74.7 75.0 69.5 71.1 DT-Fixup SQL-SP + RoBERTa (Xu et al 2021) 75.0 -70.9 -LGESQL + ELECTRA (Cao et al 2021) 75.1 -72.0 -S 2 SQL + ELECTRA (Hui et al 2022) 76.4 -72.1 -…”

Section: Results On Robustness Settingsmentioning

confidence: 96%

“…The input is one or more heterogeneous graphs (Wang et al 2020a;Hui et al 2022;Cao et al 2021;Cai et al 2021), where a node represents a question token, a table or a column, and an edge represents the relation between two nodes. Then, relation-aware transformer networks (Shaw, Uszkoreit, and Vaswani 2018) or relational graph neural networks, such as RGCN (Schlichtkrull et al 2018) and RGAT (Wang et al 2020b), are applied to encode each node.…”

Section: Graph Encodermentioning

confidence: 99%

See 1 more Smart Citation

RESDSQL: Decoupling Schema Linking and Skeleton Parsing for Text-to-SQL

Zhang

et al. 2023

AAAI

View full text Add to dashboard Cite

One of the recent best attempts at Text-to-SQL is the pre-trained language model. Due to the structural property of the SQL queries, the seq2seq model takes the responsibility of parsing both the schema items (i.e., tables and columns) and the skeleton (i.e., SQL keywords). Such coupled targets increase the difficulty of parsing the correct SQL queries especially when they involve many schema items and logic operators. This paper proposes a ranking-enhanced encoding and skeleton-aware decoding framework to decouple the schema linking and the skeleton parsing. Specifically, for a seq2seq encoder-decode model, its encoder is injected by the most relevant schema items instead of the whole unordered ones, which could alleviate the schema linking effort during SQL parsing, and its decoder first generates the skeleton and then the actual SQL query, which could implicitly constrain the SQL parsing. We evaluate our proposed framework on Spider and its three robustness variants: Spider-DK, Spider-Syn, and Spider-Realistic. The experimental results show that our framework delivers promising performance and robustness. Our code is available at https://github.com/RUCKBReasoning/RESDSQL.

show abstract

Section: Results On Robustness Settingsmentioning

confidence: 96%

Section: Graph Encodermentioning

confidence: 99%

RESDSQL: Decoupling Schema Linking and Skeleton Parsing for Text-to-SQL

Zhang

et al. 2023

AAAI

View full text Add to dashboard Cite

show abstract

“…Moreover, we train all the models on WikiBio and WikiPerson from scratch, and the training cost is rather expensive: 2.5 days using 4 NVIDIA V100 32G GPUs. Lastly, this paper does not compare the pre-trained language models (PLMs) (Devlin et al, 2019;Raffel et al, 2020;Hui et al, 2021Hui et al, , 2022, though our approach may also benefit from some pre-trained table encoders, such as TAPAS (Müller et al, 2021). The main reasons why we do not consider PLMs are that PLMs will bring an unfair comparison and bring more variables and may make our work lose focus.…”

Section: Limitationsmentioning

confidence: 99%

Plan-then-Seam: Towards Efficient Table-to-Text Generation

Li¹,

Geng²,

Fang³

et al. 2023

Preprint

View full text Add to dashboard Cite

Table -to-text generation aims at automatically generating text to help people conveniently obtain salient information in tables. Recent works explicitly decompose the generation process into content planning and surface generation stages, employing two autoregressive networks for them respectively. However, they are computationally expensive due to the nonparallelizable nature of autoregressive decoding and the redundant parameters of two networks. In this paper, we propose the first totally non-autoregressive table-to-text model (Plan-then-Seam, PTS) that produces its outputs in parallel with one single network. PTS firstly writes and calibrates one plan of the content to be generated with a novel rethinking pointer predictor, and then takes the plan as the context for seaming to decode the description. These two steps share parameters and perform iteratively to capture token interdependency while keeping parallel decoding. Experiments on two public benchmarks show that PTS achieves 3.0 ∼ 5.6 times speedup for inference time, reducing 50% parameters, while maintaining as least comparable performance against strong two-stage table-to-text competitors 1 .

show abstract

“…Research on cross-domain text-to-SQL benchmarks has led to numerous advances. Recent works (Zhao et al, 2021;Rubin and Berant, 2021;Hui et al, 2021) have achieved over 70% accuracy on Spider benchmark (Yu et al, 2018) and over 90% accuracy on WikiSQL benchmark (Zhong et al, 2017), which seems to suggest that existing models already solved most problems in this field. However, the follow-up studies from Deng et al (2021); Gan et al (2021); Suhr et al (2020); Shaw et al (2021); Oren et al (2020); Keysers et al (2020) show that the generalization performance is much worse in more challenging scenarios.…”

Section: Introductionmentioning

confidence: 99%

Exploring Underexplored Limitations of Cross-Domain Text-to-SQL Generalization

Gan¹,

Purver²

2021

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing

View full text Add to dashboard Cite

Recently, there has been significant progress in studying neural networks for translating text descriptions into SQL queries under the zeroshot cross-domain setting. Despite achieving good performance on some public benchmarks, we observe that existing text-to-SQL models do not generalize when facing domain knowledge that does not frequently appear in the training data, which may render the worse prediction performance for unseen domains. In this work, we investigate the robustness of text-to-SQL models when the questions require rarely observed domain knowledge. In particular, we define five types of domain knowledge and introduce Spider-DK (DK is the abbreviation of domain knowledge), a human-curated dataset based on the Spider benchmark for text-to-SQL translation. NL questions in Spider-DK are selected from Spider, and we modify some samples by adding domain knowledge that reflects real-world question paraphrases. We demonstrate that the prediction accuracy dramatically drops on samples that require such domain knowledge, even if the domain knowledge appears in the training set, and the model provides the correct predictions for related training samples. 1

show abstract

S2SQL: Injecting Syntax to Question-Schema Interaction Graph Encoder for Text-to-SQL Parsers

Cited by 29 publications

References 27 publications

RESDSQL: Decoupling Schema Linking and Skeleton Parsing for Text-to-SQL

RESDSQL: Decoupling Schema Linking and Skeleton Parsing for Text-to-SQL

Plan-then-Seam: Towards Efficient Table-to-Text Generation

Exploring Underexplored Limitations of Cross-Domain Text-to-SQL Generalization

Contact Info

Product

Resources

About