Upload clefip2011.py
Browse files- clefip2011.py +137 -0
clefip2011.py
ADDED
@@ -0,0 +1,137 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import pandas as pd
|
3 |
+
import datasets
|
4 |
+
|
5 |
+
|
6 |
+
class CLEFIP2011Config(datasets.BuilderConfig):
|
7 |
+
"""Custom Config for CLEFIP2011"""
|
8 |
+
|
9 |
+
def __init__(self, dataset_type=None, **kwargs):
|
10 |
+
super(CLEFIP2011Config, self).__init__(**kwargs)
|
11 |
+
self.dataset_type = dataset_type
|
12 |
+
|
13 |
+
|
14 |
+
class CLEFIP2011(datasets.GeneratorBasedBuilder):
|
15 |
+
"""Custom Dataset Loader"""
|
16 |
+
|
17 |
+
BUILDER_CONFIGS = [
|
18 |
+
CLEFIP2011Config(
|
19 |
+
name="bibliographic",
|
20 |
+
version=datasets.Version("1.0.0"),
|
21 |
+
description="CLEF-IP 2011 Bibliographic Data",
|
22 |
+
dataset_type="bibliographic",
|
23 |
+
),
|
24 |
+
]
|
25 |
+
|
26 |
+
def _info(self):
|
27 |
+
|
28 |
+
if self.config.dataset_type == "bibliographic":
|
29 |
+
features = datasets.Features(
|
30 |
+
{
|
31 |
+
"ucid": datasets.Value("string"),
|
32 |
+
"country": datasets.Value("string"),
|
33 |
+
"doc_number": datasets.Value("string"),
|
34 |
+
"kind": datasets.Value("string"),
|
35 |
+
"lang": datasets.Value("string"),
|
36 |
+
"corrected_lang": datasets.Value("string"),
|
37 |
+
"date": datasets.Value("string"),
|
38 |
+
"family_id": datasets.Value("string"),
|
39 |
+
"date_produced": datasets.Value("string"),
|
40 |
+
"status": datasets.Value("string"),
|
41 |
+
"ecla_list": datasets.Value("string"),
|
42 |
+
"applicant_name_list": datasets.Value("string"),
|
43 |
+
"inventor_name_list": datasets.Value("string"),
|
44 |
+
"title_de_text": datasets.Value("string"),
|
45 |
+
"title_fr_text": datasets.Value("string"),
|
46 |
+
"title_en_text": datasets.Value("string"),
|
47 |
+
"abstract_de_exist": datasets.Value("bool"),
|
48 |
+
"abstract_fr_exist": datasets.Value("bool"),
|
49 |
+
"abstract_en_exist": datasets.Value("bool"),
|
50 |
+
"description_de_exist": datasets.Value("bool"),
|
51 |
+
"description_fr_exist": datasets.Value("bool"),
|
52 |
+
"description_en_exist": datasets.Value("bool"),
|
53 |
+
"claims_de_exist": datasets.Value("bool"),
|
54 |
+
"claims_fr_exist": datasets.Value("bool"),
|
55 |
+
"claims_en_exist": datasets.Value("bool"),
|
56 |
+
}
|
57 |
+
)
|
58 |
+
return datasets.DatasetInfo(
|
59 |
+
description="CLEF-IP 2011 Bibliographic dataset.",
|
60 |
+
features=features,
|
61 |
+
supervised_keys=None,
|
62 |
+
)
|
63 |
+
|
64 |
+
def _split_generators(self, dl_manager):
|
65 |
+
|
66 |
+
archive_path = dl_manager.download_and_extract(
|
67 |
+
"https://huggingface.co/datasets/amylonidis/PatClass2011/resolve/main/clefip2011_bibliographic_clean.tar.gz"
|
68 |
+
)
|
69 |
+
|
70 |
+
bibliographic_file = os.path.join(archive_path, "clefip2011_bibliographic.csv")
|
71 |
+
|
72 |
+
return [
|
73 |
+
datasets.SplitGenerator(
|
74 |
+
name=datasets.Split.TRAIN,
|
75 |
+
gen_kwargs={
|
76 |
+
"filepaths": [bibliographic_file],
|
77 |
+
"split": "train",
|
78 |
+
},
|
79 |
+
),
|
80 |
+
]
|
81 |
+
|
82 |
+
def _generate_examples(self, filepaths, split):
|
83 |
+
|
84 |
+
for filepath in filepaths:
|
85 |
+
df = pd.read_csv(filepath, header=None)
|
86 |
+
|
87 |
+
column_names = [
|
88 |
+
"ucid",
|
89 |
+
"country",
|
90 |
+
"doc_number",
|
91 |
+
"kind",
|
92 |
+
"lang",
|
93 |
+
"corrected_lang",
|
94 |
+
"date",
|
95 |
+
"family_id",
|
96 |
+
"date_produced",
|
97 |
+
"status",
|
98 |
+
"ecla_list",
|
99 |
+
"applicant_name_list",
|
100 |
+
"inventor_name_list",
|
101 |
+
"title_de_text",
|
102 |
+
"title_fr_text",
|
103 |
+
"title_en_text",
|
104 |
+
"abstract_de_exist",
|
105 |
+
"abstract_fr_exist",
|
106 |
+
"abstract_en_exist",
|
107 |
+
"description_de_exist",
|
108 |
+
"description_fr_exist",
|
109 |
+
"description_en_exist",
|
110 |
+
"claims_de_exist",
|
111 |
+
"claims_fr_exist",
|
112 |
+
"claims_en_exist",
|
113 |
+
]
|
114 |
+
df.columns = column_names
|
115 |
+
|
116 |
+
df["date"] = pd.to_datetime(df["date"], format="%Y%m%d").astype(str)
|
117 |
+
|
118 |
+
df["date_produced"] = pd.to_datetime(
|
119 |
+
df["date_produced"], format="%Y%m%d"
|
120 |
+
).astype(str)
|
121 |
+
|
122 |
+
boolean_columns = [
|
123 |
+
"abstract_de_exist",
|
124 |
+
"abstract_fr_exist",
|
125 |
+
"abstract_en_exist",
|
126 |
+
"description_de_exist",
|
127 |
+
"description_fr_exist",
|
128 |
+
"description_en_exist",
|
129 |
+
"claims_de_exist",
|
130 |
+
"claims_fr_exist",
|
131 |
+
"claims_en_exist",
|
132 |
+
]
|
133 |
+
for col in boolean_columns:
|
134 |
+
df[col] = df[col].astype(bool)
|
135 |
+
|
136 |
+
for idx, row in df.iterrows():
|
137 |
+
yield idx, row.to_dict()
|