NL2SQL-BERT

Content Enhanced BERT-based Text-to-SQL Generation https://arxiv.org/abs/1910.07179

Run

1, Data prepare:

data_and_model/output_entity.py

2, Train and eval:

train.py

Results on BERT-Base-Uncased without EG

Model	Dev logical form accuracy	Dev execution accuracy	Test logical form accuracy	Test execution accuracy
SQLova	80.6	86.5	80.0	85.5
Our Methods	84.3	90.3	83.7	89.2

Data

One data look:

{
	"table_id": "1-1000181-1",
	"phase": 1,
	"question": "Tell me what the notes are for South Australia ",
	"question_tok": ["Tell", "me", "what", "the", "notes", "are", "for", "South", "Australia"],
	"sql": {
		"sel": 5,
		"conds": [
			[3, 0, "SOUTH AUSTRALIA"]
		],
		"agg": 0
	},
	"query": {
		"sel": 5,
		"conds": [
			[3, 0, "SOUTH AUSTRALIA"]
		],
		"agg": 0
	},
	"wvi_corenlp": [
		[7, 8]
	],
	"bertindex_knowledge": [0, 0, 0, 0, 4, 0, 0, 1, 3],
	"header_knowledge": [2, 0, 0, 2, 0, 1]
}

All origin data:

https://drive.google.com/file/d/1iJvsf38f16el58H4NPINQ7uzal5-V4v4

Trained model

https://drive.google.com/open?id=18MBm9qzobTBgWPZlpA2EErCQtsMhlTN2

Reference

https://github.com/naver/sqlova