[NEW]: Add dataset description
Browse files
README.md
CHANGED
@@ -10,4 +10,65 @@ tags:
|
|
10 |
pretty_name: spiderology
|
11 |
size_categories:
|
12 |
- 1K<n<10K
|
13 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
pretty_name: spiderology
|
11 |
size_categories:
|
12 |
- 1K<n<10K
|
13 |
+
---
|
14 |
+
|
15 |
+
## What is it?
|
16 |
+
* Dataset for evaluation text-to-sql models based on <a href="https://yale-lily.github.io/spider">Spider dataset</a>.
|
17 |
+
* It is divided into 32 groups **(testsets)** of samples. Each group checks some models ability.
|
18 |
+
* For each group there is control group **(simplyset)** - the same samples, but without considering features.
|
19 |
+
* This is very similar to A/B-testing technic.
|
20 |
+
* All the simplified samples (from control group) are new.
|
21 |
+
* Test samples contains:
|
22 |
+
1. Examples extracted from Spider (not only from "dev", but also from "train" and "train_other") **Be careful,** if you use it for training!
|
23 |
+
2. Examples extracted from Spider, but corrected from ambiguity and errors
|
24 |
+
3. Manually created (new) examples
|
25 |
+
|
26 |
+
## Structure
|
27 |
+
### Folders and files
|
28 |
+
* [\[all\]](/all) - the hole dataset in one place
|
29 |
+
* ```all/testset_all.json``` - the samples with features from all testsets
|
30 |
+
* ```all/simplyset_all.json``` - the samples without features from all testsets
|
31 |
+
* ```all/spiderology.json``` - the samples with features from all testsets and simplysets
|
32 |
+
* ```all/tables_all.json``` - the schemes of all tables required for requests
|
33 |
+
* ```all/databases``` - the SQLite-databases for all requests in testsets and simplysets
|
34 |
+
* [\[testsets\]](/testsets) - folder with testsets calledby their name
|
35 |
+
* ```testsets/X/testset_X.json``` - the samples with feature X. Structure of samples is below
|
36 |
+
* ```testsets/X/simplyset_X.json``` - the samples from testset without feature X. Structure of samples is the same
|
37 |
+
* ```testsets/X/tables_X.json``` - the schemes of all tables for testset and simplyset
|
38 |
+
* ```testsets/X/databases_X.json``` - the SQLite-databases for all requests in testsets and simplysets X
|
39 |
+
* ```testsets/X/markup_X.json``` - the simplysets and testsets in one table (in the form convenient for comparing)
|
40 |
+
* [\[experiments/predictions\]](/experiments/predictions) - folder with testsets calledby their name
|
41 |
+
* ```/experiments/predictions/spiderology_with_predictions.json``` - all samples with predictions of two considering models
|
42 |
+
|
43 |
+
### Structure of testsets and simplysets
|
44 |
+
|
45 |
+
```
|
46 |
+
{
|
47 |
+
"id": Unique ID of sample
|
48 |
+
"db_id": Name od database, corresponds to folder with DB in the databases_X
|
49 |
+
"query": SQL-query
|
50 |
+
"query_toks": Tokens of SQL-query
|
51 |
+
"query_toks_no_value": Tokens of SQL-query, filter value replaced by token 'value'
|
52 |
+
"question": Natural language question
|
53 |
+
"question_toks": Tokens of natural language question
|
54 |
+
"sql": Structured SQL-query (in Spider format
|
55 |
+
"tags": List of tags (features an sub-characteristics)
|
56 |
+
"source": Variant of sources (see below)
|
57 |
+
"parents_id": None for samples in testset or ID of simplified sample for simplification
|
58 |
+
"is_simplification": 'true' for simplifications, 'false' for other samples
|
59 |
+
"simplifications_tags": None for samples from testset, list of named simplified features for simplyset,
|
60 |
+
"predictions": Dictionary with SQL-queries predicted by models (exists only in the folder "experiments"),
|
61 |
+
"evaluation": Results of exact_matching, execution and human evaluation (exists only in the folder "experiments")
|
62 |
+
}
|
63 |
+
```
|
64 |
+
|
65 |
+
#### Sources
|
66 |
+
* ```SIMPLIFIED``` = Simplification of samples
|
67 |
+
* ```SPIDER_DEV``` = Samples from Spider test set ('dev')
|
68 |
+
* ```SPIDER_TRAIN``` = Samples from Spider train set ('train')
|
69 |
+
* ```SPIDER_TRAIN_OTHERS``` = Samples from Spider train set ('train_other')
|
70 |
+
* ```ADDED``` = Created samples with features (not from Spider)
|
71 |
+
|
72 |
+
|
73 |
+
## Test-set description
|
74 |
+
Here: https://github.com/rology/spiderology?tab=readme-ov-file#test-set-description
|