Datasets:

Languages:
Japanese
License:

The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.

Dataset Card for CAMERA 📷

Dataset Summary

From the official README.md:

CAMERA (CyberAgent Multimodal Evaluation for Ad Text GeneRAtion) is the Japanese ad text generation dataset. We hope that our dataset will be useful in research for realizing more advanced ad text generation models.

The data is distributed through CyberAgent's official Hugging Face release. The creative-graphic-design/CAMERA Hub repository hosts this card only and contains no data files. This repository's loader can no longer rebuild the data because its source archives return HTTP 404.

Supported Tasks and Leaderboards

[More Information Needed]

Supported Tasks

[More Information Needed]

Leaderboard

[More Information Needed]

Languages

The language data in CAMERA is in Japanese (BCP-47 ja-JP).

Dataset Structure

Data Instances

Load a specific configuration by passing its name to load_dataset:

without-lp-images

import datasets as ds

dataset = ds.load_dataset("cyberagent/camera", name="without-lp-images")

print(dataset)
# DatasetDict({
#     test: Dataset({
#         features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation'],
#         num_rows: 872
#     })
#     dev: Dataset({
#         features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation'],
#         num_rows: 3098
#     })
#     train: Dataset({
#         features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation'],
#         num_rows: 12395
#     })
# })

An example of the CAMERA (w/o LP images) dataset looks as follows:

{
  "asset_id": 13861,
  "kw": "仙台 ホテル",
  "lp_meta_description": "仙台のホテルや旅館をお探しなら楽天トラベルへ!楽天ポイントが使えて、貯まって、とってもお得な宿泊予約サイトです。さらに割引クーポンも使える!国内ツアー・航空券・レンタカー・バス予約も!",
  "title_org": "仙台市のホテル",
  "title_ne1": "",
  "title_ne2": "",
  "title_ne3": "",
  "domain": "",
  "parsed_full_text_annotation": {
    "text": [
      "trivago",
      "Oops...AccessDenied 可",
      "Youarenotallowedtoviewthispage!Ifyouthinkthisisanerror,pleasecontacttrivago.",
      "Errorcode:0.3c99e86e.1672026945.25ba640YourIP:240d:1a:4d8:2800:b9b0:ea86:2087:d141AffectedURL:https://www.trivago.jp/ja/odr/%E8%BB%92",
      "%E4%BB%99%E5%8F%B0-%E5%9B%BD%E5%86%85?search=20072325",
      "Backtotrivago"
    ],
    "xmax": [653, 838, 765, 773, 815, 649],
    "xmin": [547, 357, 433, 420, 378, 550],
    "ymax": [47, 390, 475, 558, 598, 663],
    "ymin": [18, 198, 439, 504, 566, 651]
  }
}

with-lp-images

import datasets as ds

dataset = ds.load_dataset("cyberagent/camera", name="with-lp-images")

print(dataset)
# DatasetDict({
#     test: Dataset({
#         features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation', 'lp_image'],
#         num_rows: 872
#     })
#     dev: Dataset({
#         features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation', 'lp_image'],
#         num_rows: 3098
#     })
#     train: Dataset({
#         features: ['asset_id', 'kw', 'lp_meta_description', 'title_org', 'title_ne1', 'title_ne2', 'title_ne3', 'domain', 'parsed_full_text_annotation', 'lp_image'],
#         num_rows: 12395
#     })
# })

An example of the CAMERA (w/ LP images) dataset looks as follows:

{
    "asset_id": 13861,
    "kw": "仙台 ホテル",
    "lp_meta_description": "仙台のホテルや旅館をお探しなら楽天トラベルへ!楽天ポイントが使えて、貯まって、とってもお得な宿泊予約サイトです。さらに割引クーポンも使える!国内ツアー・航空券・レンタカー・バス予約も!",
    "title_org": "仙台市のホテル",
    "title_ne1": "",
    "title_ne2": "",
    "title_ne3": "",
    "domain": "",
    "parsed_full_text_annotation": {
        "text": [
            "trivago",
            "Oops...AccessDenied 可",
            "Youarenotallowedtoviewthispage!Ifyouthinkthisisanerror,pleasecontacttrivago.",
            "Errorcode:0.3c99e86e.1672026945.25ba640YourIP:240d:1a:4d8:2800:b9b0:ea86:2087:d141AffectedURL:https://www.trivago.jp/ja/odr/%E8%BB%92", "%E4%BB%99%E5%8F%B0-%E5%9B%BD%E5%86%85?search=20072325",
            "Backtotrivago"
        ],
        "xmax": [
            653,
            838,
            765,
            773,
            815,
            649
        ],
        "xmin": [
            547,
            357,
            433,
            420,
            378,
            550
        ],
        "ymax": [
            47,
            390,
            475,
            558,
            598,
            663
        ],
        "ymin": [
            18,
            198,
            439,
            504,
            566,
            651
        ]
    },
    "lp_image": <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=1200x680 at 0x7F8513446B20>
}

Data Fields

without-lp-images

  • asset_id: ids (associated with LP images)
  • kw: search keyword
  • lp_meta_description: meta description extracted from LP (i.e., LP Text)
  • title_org: ad text (original gold reference)
  • title_ne{1-3}: ad text (additonal gold references for multi-reference evaluation)
  • domain: industry domain (HR, EC, Fin, Edu) for industry-wise evaluation
  • parsed_full_text_annotation: OCR results for LP images

with-lp-images

  • asset_id: ids (associated with LP images)
  • kw: search keyword
  • lp_meta_description: meta description extracted from LP (i.e., LP Text)
  • title_org: ad text (original gold reference)
  • title_ne{1-3}: ad text (additional gold references for multi-reference evaluation)
  • domain: industry domain (HR, EC, Fin, Edu) for industry-wise evaluation
  • parsed_full_text_annotation: OCR results for LP images
  • lp_image: Landing page (LP) image

Data Splits

The split names and row counts below match both configurations in the official release card's dataset_info. The reference counts and industry-domain labels follow the official paper. The with-lp-images configuration adds the lp_image feature.

Split # of data # of reference ad text industry domain label
Train 12,395 1 -
Dev 3,098 1 -
Test 872 4 ✔

The paper reports 869 test examples; the official release contains 872.

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

[More Information Needed]

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

[More Information Needed]

Dataset Curators

[More Information Needed]

Licensing Information

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Citation Information

@inproceedings{mita-et-al:nlp2023,
    author =    "三田 雅人 and 村上 聡一朗 and 張 培楠",
    title =	    "広告文生成タスクの規定とベンチマーク構築",
    booktitle = "言語処理学会 第 29 回年次大会",
    year =      2023,
}

Contributions

Thanks to Masato Mita, Soichiro Murakami, and Peinan Zhang for creating this dataset.

Downloads last month
96

Models trained or fine-tuned on creative-graphic-design/CAMERA