Datasets:
config and copied code from cnn/dailymail
Browse files- citesum.py +129 -0
citesum.py
ADDED
@@ -0,0 +1,129 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2020 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
|
3 |
+
#
|
4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
5 |
+
# you may not use this file except in compliance with the License.
|
6 |
+
# You may obtain a copy of the License at
|
7 |
+
#
|
8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
9 |
+
#
|
10 |
+
# Unless required by applicable law or agreed to in writing, software
|
11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
13 |
+
# See the License for the specific language governing permissions and
|
14 |
+
# limitations under the License.
|
15 |
+
|
16 |
+
# Lint as: python3
|
17 |
+
"""CiteSum dataset"""
|
18 |
+
|
19 |
+
import hashlib
|
20 |
+
import os
|
21 |
+
|
22 |
+
import datasets
|
23 |
+
|
24 |
+
|
25 |
+
logger = datasets.logging.get_logger(__name__)
|
26 |
+
|
27 |
+
|
28 |
+
_HOMEPAGE = "https://github.com/morningmoni/CiteSum"
|
29 |
+
|
30 |
+
_DESCRIPTION = """\
|
31 |
+
Citation Text-guided Scientific Extreme Summarization and Low-resource Domain Adaptation
|
32 |
+
CiteSum contains TLDR summaries for scientific papers from their citation texts without human annotation.
|
33 |
+
CiteSum is around 30 times larger than the previous human-curated dataset SciTLDR.
|
34 |
+
"""
|
35 |
+
|
36 |
+
# The second citation introduces the source data, while the first
|
37 |
+
# introduces the specific form (non-anonymized) we use here.
|
38 |
+
_CITATION = """\
|
39 |
+
@misc{https://doi.org/10.48550/arxiv.2205.06207,
|
40 |
+
doi = {10.48550/ARXIV.2205.06207},
|
41 |
+
url = {https://arxiv.org/abs/2205.06207},
|
42 |
+
author = {Mao, Yuning and Zhong, Ming and Han, Jiawei},
|
43 |
+
keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
|
44 |
+
title = {CiteSum: Citation Text-guided Scientific Extreme Summarization and Low-resource Domain Adaptation},
|
45 |
+
publisher = {arXiv},
|
46 |
+
year = {2022},
|
47 |
+
copyright = {Creative Commons Attribution 4.0 International}
|
48 |
+
}
|
49 |
+
|
50 |
+
"""
|
51 |
+
|
52 |
+
_DOWNLOAD_URL = "https://drive.google.com/file/d/1ndHCREXGSPnDUNllladh9qCtayqbXAfJ"
|
53 |
+
|
54 |
+
|
55 |
+
class CiteSumConfig(datasets.BuilderConfig):
|
56 |
+
"""BuilderConfig for CiteSum."""
|
57 |
+
|
58 |
+
def __init__(self, **kwargs):
|
59 |
+
"""BuilderConfig for CiteSum.
|
60 |
+
Args:
|
61 |
+
**kwargs: keyword arguments forwarded to super.
|
62 |
+
"""
|
63 |
+
super().__init__(**kwargs)
|
64 |
+
|
65 |
+
|
66 |
+
class CiteSum(datasets.GeneratorBasedBuilder):
|
67 |
+
"""CiteSum summarization dataset."""
|
68 |
+
|
69 |
+
BUILDER_CONFIGS = [CiteSumConfig(name="citesum", description="Plain text")]
|
70 |
+
|
71 |
+
def _info(self):
|
72 |
+
return datasets.DatasetInfo(
|
73 |
+
description=_DESCRIPTION,
|
74 |
+
features=datasets.Features(
|
75 |
+
{
|
76 |
+
"src": datasets.Value("string"),
|
77 |
+
"tgt": datasets.Value("string"),
|
78 |
+
"paper_id": datasets.Value("string"),
|
79 |
+
"title": datasets.Value("string"),
|
80 |
+
"discipline": {
|
81 |
+
"venue": datasets.Value("string"),
|
82 |
+
"journal": datasets.Value("string"),
|
83 |
+
"mag_field_of_study": datasets.features.Sequence(
|
84 |
+
datasets.Value("string")
|
85 |
+
),
|
86 |
+
},
|
87 |
+
}
|
88 |
+
),
|
89 |
+
supervised_keys=None,
|
90 |
+
homepage=_HOMEPAGE,
|
91 |
+
citation=_CITATION,
|
92 |
+
)
|
93 |
+
|
94 |
+
def _split_generators(self, dl_manager):
|
95 |
+
dl_paths = dl_manager.download(_DOWNLOAD_URL)
|
96 |
+
return [
|
97 |
+
datasets.SplitGenerator(
|
98 |
+
name=split,
|
99 |
+
gen_kwargs={
|
100 |
+
"urls_file": dl_paths[split],
|
101 |
+
"files_per_archive": [
|
102 |
+
dl_manager.iter_archive(dl_paths["cnn_stories"]),
|
103 |
+
dl_manager.iter_archive(dl_paths["dm_stories"]),
|
104 |
+
],
|
105 |
+
},
|
106 |
+
)
|
107 |
+
for split in [
|
108 |
+
datasets.Split.TRAIN,
|
109 |
+
datasets.Split.VALIDATION,
|
110 |
+
datasets.Split.TEST,
|
111 |
+
]
|
112 |
+
]
|
113 |
+
|
114 |
+
def _generate_examples(self, urls_file, files_per_archive):
|
115 |
+
urls = _get_url_hashes(urls_file)
|
116 |
+
idx = 0
|
117 |
+
for files in files_per_archive:
|
118 |
+
for path, file in files:
|
119 |
+
hash_from_path = _get_hash_from_path(path)
|
120 |
+
if hash_from_path in urls:
|
121 |
+
article, highlights = _get_art_abs(file, self.config.version)
|
122 |
+
if not article or not highlights:
|
123 |
+
continue
|
124 |
+
yield idx, {
|
125 |
+
_ARTICLE: article,
|
126 |
+
_HIGHLIGHTS: highlights,
|
127 |
+
"id": hash_from_path,
|
128 |
+
}
|
129 |
+
idx += 1
|