Towards Complex Text-to-SQL in Cross-Domain Database with Intermediate Representation

Guo, Jiaqi; Zhan, Zecheng; Gao, Yan; Xiao, Yan; Lou, Jian–Guang; Liu, Ting; Zhang, Dongmei

doi:10.48550/arxiv.1905.08205

Cited by 23 publications

(40 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…EditSQL uses an editing mechanism to introduce historical information for user queries, and its matching accuracy on Spider dataset reaches up to 32.9. IRNet (Guo et al, 2019) adopts an intermediate representation named SemQL to translate complex SQL queries into a syntax tree. Using pointer network (Vinyals et al, 2015) for downstream tasks, it achieves an accuracy of 54.7 on Spider test set.…”

Section: Related Workmentioning

confidence: 99%

“…This type of method requires manual summarization of experience and has a high time cost. In addition, with the switch of application scenarios, the existing templates are often difficult to meet the requirements, and the migration is poor; (2) Based on the deep learning method, the neural network is used for end-to-end implementation (Zhong et al, 2017, Yu et al, 2018a,b, Bogin et al, 2019, Guo et al, 2019. This method can be self-optimized by continuously adding sample information.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

GP: Context-free Grammar Pre-training for Text-to-SQL Parsers

Zhao,

Cao,

Zhao

2021

Preprint

View full text Add to dashboard Cite

A new method for Text-to-SQL parsing, Grammar Pre-training (GP), is proposed to decode deep relations between question and database. Firstly, to better utilize the information of databases, a random value is added behind a question word which is recognized as a column, and the new sentence serves as the model input. Secondly, initialization of vectors for decoder part is optimized, with reference to the former encoding so that question information can be concerned. Finally, a new approach called flooding level is adopted to get the non-zero training loss which can generalize better results. By encoding the sentence with GRAPPA and RAT-SQL model, we achieve better performance on spider, a cross-DB 69.8 test). Experiments show that our method is easier to converge during training and has excellent robustness.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

GP: Context-free Grammar Pre-training for Text-to-SQL Parsers

Zhao,

Cao,

Zhao

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…The recent state-of-the-art models evaluated on Spider use various attentional architectures for question/schema encoding and AST-based structural architectures for query decoding. IRNet (Guo et al, 2019) encodes the question and schema separately with LSTM and self-attention respectively, augmenting them with custom type vectors for schema linking. They further use the AST-based decoder of Yin and Neubig (2017) to decode a query in an intermediate representation (IR) that exhibits higher-level abstraction structure than SQL.…”

Section: Related Workmentioning

confidence: 99%

“…This procedure matches that ofGuo et al (2019), but we use the matching information differently in RAT.…”

mentioning

confidence: 99%

RAT-SQL: Relation-Aware Schema Encoding and Linking for Text-to-SQL Parsers

Wang¹,

Shin²,

Liu³

et al. 2019

Preprint

View full text Add to dashboard Cite

When translating natural language questions into SQL queries to answer questions from a database, contemporary semantic parsing models struggle to generalize to unseen database schemas. The generalization challenge lies in (a) encoding the database relations in an accessible way for the semantic parser, and (b) modeling alignment between database columns and their mentions in a given query. We present a unified framework, based on the relation-aware self-attention mechanism, to address schema encoding, schema linking, and feature representation within a text-to-SQL encoder. On the challenging Spider dataset this framework boosts the exact match accuracy to 53.7%, compared to 47.4% for the stateof-the-art model unaugmented with BERT embeddings. In addition, we observe qualitative improvements in the model's understanding of schema linking and alignment.

show abstract

“…Tables store rich numerical data, and a wide range of recent table-related tasks highly rely on numerical reasoning, such as spreadsheet formula prediction (Chen et al, 2021a), table structure understanding (Koci et al, 2019), question answering over tables (Chen et al, 2021b;Cheng et al, 2021), and data-to-text (Suadaa et al, 2021;Cheng et al, 2021). To achieve better numerical reasoning capabilities, some works explore to design domain specific languages (Chen et al, 2020(Chen et al, , 2021bCheng et al, 2021;Guo et al, 2019) and inject preexecuted operations (Suadaa et al, 2021), but are task-specific and suffer from shortage of labels.…”

Section: Introductionmentioning

confidence: 99%

FORTAP: Using Formulas for Numerical-Reasoning-Aware Table Pretraining

Cheng¹,

Dong²,

Jia³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Tables store rich numerical data, but numerical reasoning over tables is still a challenge. In this paper, we find that the spreadsheet formula, which performs calculations on numerical values in tables, is naturally a strong supervision of numerical reasoning. More importantly, large amounts of spreadsheets with expert-made formulae are available on the web and can be obtained easily. FORTAP is the first method for numerical-reasoning-aware table pretraining by leveraging large corpus of spreadsheet formulae. We design two formula pretraining tasks to explicitly guide FORTAP to learn numerical reference and calculation in semi-structured tables. FORTAP achieves state-of-the-art results on two representative downstream tasks, cell type classification and formula prediction, showing great potential of numerical-reasoning-aware pretraining.

show abstract

Towards Complex Text-to-SQL in Cross-Domain Database with Intermediate Representation

Cited by 23 publications

References 34 publications

GP: Context-free Grammar Pre-training for Text-to-SQL Parsers

GP: Context-free Grammar Pre-training for Text-to-SQL Parsers

RAT-SQL: Relation-Aware Schema Encoding and Linking for Text-to-SQL Parsers

FORTAP: Using Formulas for Numerical-Reasoning-Aware Table Pretraining

Contact Info

Product

Resources

About