Global Reasoning over Database Structures for Text-to-SQL Parsing

@article{Bogin2019GlobalRO,
  title={Global Reasoning over Database Structures for Text-to-SQL Parsing},
  author={Ben Bogin and Matt Gardner and Jonathan Berant},
  journal={ArXiv},
  year={2019},
  volume={abs/1908.11214}
}
  • Ben Bogin, Matt Gardner, Jonathan Berant
  • Published in IJCNLP 2019
State-of-the-art semantic parsers rely on auto-regressive decoding, emitting one symbol at a time. When tested against complex databases that are unobserved at training time (zero-shot), the parser often struggles to select the correct set of database constants in the new database, due to the local nature of decoding. In this work, we propose a semantic parser that globally reasons about the structure of the output query to make a more contextually-informed selection of database constants. We… CONTINUE READING

Figures, Tables, Results, and Topics from this paper.

Key Quantitative Results

  • We show that both our contributions improve performance, leading to an accuracy of 47.4%, well beyond the current state-of-the-art of 39.4%.

References

Publications referenced by this paper.
SHOWING 1-10 OF 19 REFERENCES

SyntaxSQLNet: Syntax Tree Networks for Complex and Cross-DomainText-to-SQL Task

Tao Yu, Michihiro Yasunaga, +4 authors Dragomir R. Radev
  • EMNLP
  • 2018
VIEW 5 EXCERPTS
HIGHLY INFLUENTIAL

We add cell values to the graph, in a similar fashion to Krishnamurthy et al

A Re-implementation We perform a simple modification Bogin
  • (2017). Specifically, we extract the cells of the first 5000 rows of
  • 2019