Kosuke-Yamada commited on
Commit
031b3a0
1 Parent(s): b2e46ed

update readme.md

Browse files
Files changed (1) hide show
  1. README.md +167 -13
README.md CHANGED
@@ -1,4 +1,13 @@
1
  ---
 
 
 
 
 
 
 
 
 
2
  dataset_info:
3
  - config_name: with-lp-images
4
  features:
@@ -104,16 +113,161 @@ configs:
104
  path: without-lp-images/validation-*
105
  - split: test
106
  path: without-lp-images/test-*
107
- license: cc-by-nc-sa-4.0
108
- language: ja
109
- tags:
110
- - e-commerce
111
- - education
112
- - finance
113
- - hr
114
- task_categories:
115
- - text2text-generation
116
- - image-to-text
117
- size_categories: 10K<n<100K
118
- pretty_name: camera
119
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ license: cc-by-nc-sa-4.0
3
+ language: ja
4
+ tags:
5
+ - advertisement
6
+ task_categories:
7
+ - text2text-generation
8
+ - image-to-text
9
+ size_categories: 10K<n<100K
10
+ pretty_name: camera
11
  dataset_info:
12
  - config_name: with-lp-images
13
  features:
 
113
  path: without-lp-images/validation-*
114
  - split: test
115
  path: without-lp-images/test-*
116
+ ---
117
+
118
+ # Dataset Card for CAMERA📷:
119
+
120
+ ## Table of Contents:
121
+ - [Dataset Card for Camera](#dataset-card-for-camera)
122
+ - [Table of Contents](#table-of-contents)
123
+ - [Dataset Details](#dataset-details)
124
+ - [Dataset Description](#dataset-description)
125
+ - [Dataset Sources](#dataset-sources)
126
+ - [Uses](#uses)
127
+ - [Direct Use](#direct-use)
128
+ - [Dataset Information](#datasest-information)
129
+ - [Data Example](#data-example)
130
+ - [Dataset Structure](#dataset-structure)
131
+ - [Citation](#citation)
132
+
133
+
134
+ ## Dataset Details
135
+
136
+ ### Dataset Description
137
+ CAMERA (CyberAgent Multimodal Evaluation for Ad Text GeneRAtion) is the Japanese ad text generation dataset.
138
+
139
+ ### Dataset Sources
140
+
141
+ - **Homepage:** [Github](https://github.com/CyberAgentAILab/camera)
142
+ - **Paper:** [Striking Gold in Advertising: Standardization and Exploration of Ad Text
143
+ Generation](https://arxiv.org/abs/2309.12030)
144
+
145
+ ## Uses
146
+
147
+ ### Direct Use
148
+
149
+ - Dataset with lp images (with-lp-images)
150
+ ```python
151
+ import datasets
152
+ dataset = datasets.load_dataset("cyberagent/camera", name="with-lp-images")
153
+ ```
154
+
155
+ - Dataset without lp images (without-lp-images)
156
+ ```python
157
+ import datasets
158
+ dataset = datasets.load_dataset("cyberagent/camera", name="without-lp-images")
159
+ ```
160
+
161
+ ### Dataset Information
162
+
163
+ - with-lp-images
164
+ ```
165
+ DatasetDict({
166
+ train: Dataset({
167
+ features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation', 'lp_image'],
168
+ num_rows: 12395
169
+ })
170
+ validation: Dataset({
171
+ features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation', 'lp_image'],
172
+ num_rows: 3098
173
+ })
174
+ test: Dataset({
175
+ features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation', 'lp_image'],
176
+ num_rows: 872
177
+ })
178
+ })
179
+ ```
180
+
181
+ - without-lp-images
182
+ ```
183
+ DatasetDict({
184
+ train: Dataset({
185
+ features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation'],
186
+ num_rows: 12395
187
+ })
188
+ validation: Dataset({
189
+ features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation'],
190
+ num_rows: 3098
191
+ })
192
+ test: Dataset({
193
+ features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation'],
194
+ num_rows: 872
195
+ })
196
+ })
197
+ ```
198
+
199
+ ### Data Example
200
+
201
+ ```
202
+ {'asset_id': 6041,
203
+ 'kw': 'GLLARE MARUYAMA',
204
+ 'lp_meta_description': '美容サロン ブルーヘアー 札幌市 西区 琴似 創業34年 かゆみ、かぶれを防ぎ、美しい髪へ',
205
+ 'title_org': '北海道、水の教会で結婚式',
206
+ 'title_ne1': '',
207
+ 'title_ne2': '',
208
+ 'title_ne3': '',
209
+ 'domain': '',
210
+ 'parsed_full_text_annotation': {
211
+ 'text': ['表参道',
212
+ '名古屋',
213
+ '梅田',
214
+ ...
215
+ '成約者様専用ページ',
216
+ '個人情報保護方針',
217
+ '星野リゾートトマム'],
218
+ 'xmax': [163,
219
+ 162,
220
+ 157,
221
+ ...
222
+ 1047,
223
+ 1035,
224
+ 1138],
225
+ 'xmin': [125,
226
+ 125,
227
+ 129,
228
+ ...
229
+ 937,
230
+ 936,
231
+ 1027],
232
+ 'ymax': [9652,
233
+ 9791,
234
+ 9928,
235
+ ...
236
+ 17119,
237
+ 17154,
238
+ 17515],
239
+ 'ymin': [9642,
240
+ 9781,
241
+ 9918,
242
+ ...
243
+ 17110,
244
+ 17143,
245
+ 17458]},
246
+ 'lp_image': <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=1200x17596>}
247
+ ```
248
+
249
+ ### Dataset Structure
250
+
251
+ | Name | Description |
252
+ | ---- | ---- |
253
+ | asset_id | ids (associated with LP images) |
254
+ | kw | search keyword |
255
+ | lp_meta_description | meta description extracted from LP (i.e., LP Text)|
256
+ | title_org | ad text (original gold reference) |
257
+ | title_ne{1-3} | ad text (additonal gold references for multi-reference evaluation |
258
+ | domain | industry domain (HR, EC, Fin, Edu) for industry-wise evaluation |
259
+ | parsed_full_text_annotation | OCR result for LP image |
260
+ | lp_image | LP image |
261
+
262
+ ## Citation
263
+
264
+ ```
265
+ @misc{mita2024striking,
266
+ title={Striking Gold in Advertising: Standardization and Exploration of Ad Text Generation},
267
+ author={Masato Mita and Soichiro Murakami and Akihiko Kato and Peinan Zhang},
268
+ year={2024},
269
+ eprint={2309.12030},
270
+ archivePrefix={arXiv},
271
+ primaryClass={id='cs.CL' full_name='Computation and Language' is_active=True alt_name='cmp-lg' in_archive='cs' is_general=False description='Covers natural language processing. Roughly includes material in ACM Subject Class I.2.7. Note that work on artificial languages (programming languages, logics, formal systems) that does not explicitly address natural-language issues broadly construed (natural-language processing, computational linguistics, speech, text retrieval, etc.) is not appropriate for this area.'}
272
+ }
273
+ ```