kasnerz commited on
Commit
cac4970
1 Parent(s): f421c01

Delete _charttotext-s.py

Browse files
Files changed (1) hide show
  1. _charttotext-s.py +0 -117
_charttotext-s.py DELETED
@@ -1,117 +0,0 @@
1
- #!/usr/bin/env python3
2
-
3
- """
4
- The script used to load the dataset from the original source.
5
- """
6
-
7
- import json
8
- import datasets
9
- import os
10
- import csv
11
-
12
- _CITATION = """\
13
- @inproceedings{kantharaj2022chart,
14
- title={Chart-to-Text: A Large-Scale Benchmark for Chart Summarization},
15
- author={Kantharaj, Shankar and Leong, Rixie Tiffany and Lin, Xiang and Masry, Ahmed and Thakkar, Megh and Hoque, Enamul and Joty, Shafiq},
16
- booktitle={Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
17
- pages={4005--4023},
18
- year={2022}
19
- }
20
- """
21
- _DESCRIPTION = """\
22
- Chart-to-Text is a large-scale benchmark with two datasets and a total of 44,096 charts covering a wide range of topics and chart types.
23
- This dataset CONTAINS ONLY the Statista subset from the benchmark.
24
- Statista (statista.com) is an online platform that regularly publishes charts on a wide range of topics including economics, market and opinion research.
25
-
26
- Statistics:
27
- Total charts: 27868
28
-
29
- === Chart Type Information ===
30
- Number of charts of each chart type
31
- column: 16319
32
- bar: 8272
33
- line: 2646
34
- pie: 408
35
- table: 223
36
-
37
- === Token Information ===
38
- Average token count per summary: 53.65027989091431
39
- Total tokens: 1495126
40
- Total types (unique tokens): 39598
41
- === Sentence Information ===
42
- Average sentence count per summary: 2.5596741782689825
43
- """
44
-
45
- _URL = "https://github.com/vis-nlp/Chart-to-text/tree/main/statista_dataset/dataset"
46
- _LICENSE = "GNU General Public License v3.0"
47
-
48
- class ChartToTextS(datasets.GeneratorBasedBuilder):
49
- VERSION = "1.0.0"
50
-
51
- def _info(self):
52
- return datasets.DatasetInfo(
53
- description=_DESCRIPTION,
54
- features=datasets.Features({
55
- 'title' : datasets.Value(dtype='string'),
56
- 'ref' : datasets.Value(dtype='string'),
57
- 'content' : datasets.Value(dtype='large_string'),
58
- }),
59
- supervised_keys=None,
60
- homepage=_URL,
61
- citation=_CITATION,
62
- license=_LICENSE,
63
- )
64
-
65
- def _split_generators(self, dl_manager):
66
- """Returns SplitGenerators."""
67
- return [
68
- datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": "dataset", "split" : "train"}),
69
- datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": "dataset", "split" : "dev"}),
70
- datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": "dataset", "split" : "test"}),
71
- ]
72
-
73
- def _generate_examples(self, filepath, split):
74
- data = []
75
- mapping_file = split if split != "dev" else "val"
76
-
77
- with open(
78
- os.path.join(
79
- filepath, "dataset_split", f"{mapping_file}_index_mapping.csv"
80
- )
81
- ) as f:
82
- next(f)
83
- for i, line in enumerate(f):
84
- subdir = "." if line.startswith("two_col") else "multiColumn"
85
- filename = line.split("-")[1].split(".")[0]
86
-
87
- with open(
88
- os.path.join(filepath, subdir, "data", filename + ".csv")
89
- ) as g:
90
- content = []
91
- reader = csv.reader(g, delimiter=",", quotechar='"')
92
- for row in reader:
93
- content.append(row)
94
-
95
- with open(
96
- os.path.join(filepath, subdir, "captions", filename + ".txt")
97
- ) as g:
98
- ref = g.read().rstrip("\n")
99
-
100
- with open(
101
- os.path.join(filepath, subdir, "titles", filename + ".txt")
102
- ) as g:
103
- title = g.read().rstrip("\n")
104
-
105
- data.append(
106
- {"content": content, "ref": ref, "title": title}
107
- )
108
-
109
- if i % 1000 == 0:
110
- print(f"Loaded {i} items")
111
-
112
- for example_idx, entry in enumerate(data):
113
- yield example_idx, {key: str(value) for key, value in entry.items()}
114
-
115
- if __name__ == '__main__':
116
- dataset = datasets.load_dataset(__file__)
117
- dataset.push_to_hub("kasnerz/charttotext-s")