metadata

annotations_creators:
  - found
language:
  - ru
license:
  - cc0-1.0
multilinguality:
  - monolingual
pretty_name: Grustnogram Social Media Dataset
size_categories:
  - 100K<n<1M
source_datasets:
  - original
task_categories:
  - image-classification
  - image-to-text
  - text-classification
task_ids:
  - multi-label-image-classification
  - image-captioning
  - sentiment-analysis
configs:
  - config_name: train
    data_files:
      - split: posts
        path: posts.jsonl.zst
      - split: texts
        path: texts.jsonl.zst
    default: true
  - config_name: images
    data_files:
      - split: media
        path: media/*
      - split: avatar
        path: avatar/*

Dataset Card for Grustnogram

Dataset Summary

This dataset contains 597,704 posts from Grustnogram.ru, a Russian "emotional network" similar to Instagram but with a distinctive black and white filter aesthetic and dark atmosphere. The dataset includes 542,917 image posts with associated metadata and 54,787 anonymous text-only posts.

Languages

The dataset is primarily in Russian (ru).

Dataset Structure

Data Splits

The dataset is divided into two main splits:

posts: 542,917 entries of image posts with user information
texts: 54,787 entries of anonymous text-only posts

Data Fields

posts

id: Unique identifier for the post (integer)
id_user: Unique identifier for the user who created the post (integer)
id_circle: Circle identifier (integer)
circle_only: Flag indicating if post is limited to a circle (integer)
url: Short URL identifier for the post (string)
media: Array of image URLs (array of strings)
media_250: Array of thumbnail image URLs at 250px (array of strings)
media_100: Array of thumbnail image URLs at 100px (array of strings)
sound_duration: Duration of attached sound if present (integer)
text: Caption or description of the post (string)
filter: Filter applied to the image (integer)
w: Width of the image (integer)
h: Height of the image (integer)
links: Array of embedded links (array)
likes: Object containing like count and user information (object)
comments: Object containing comments data and count (object)
created_at: Unix timestamp when the post was created (integer)
edited: Flag indicating if the post was edited (integer)
repost: Flag indicating if the post is a repost (integer)
type: Type of content (string)
user: Object containing user information including:
- id: User ID (integer)
- nickname: Username (string)
- avatar: URL to user's avatar (string)
- avatar_250: URL to 250px thumbnail of user's avatar (string)
- avatar_100: URL to 100px thumbnail of user's avatar (string)
- poet: Flag for poet status (integer)
- fire_till: Fire status expiration (integer)
- have_fire: Flag for fire status (integer)
liked: Flag indicating if the viewer liked the post (integer)
favorite: Flag indicating if the viewer favorited the post (integer)
canDelete: Flag indicating if the post can be deleted by viewer (integer)
canFollow: Flag indicating if the user can be followed by viewer (integer)
canBlockUser: Flag indicating if the user can be blocked by viewer (integer)
canEdit: Flag indicating if the post can be edited by viewer (integer)

texts

id: Unique identifier for the text post (integer)
url: Short URL identifier for the post (string)
secret: Text content of the post (string)
created_at: Unix timestamp when the post was created (integer)
likes: Object containing like count (object)
comments: Object containing comments data and count (object)
tags: Array of tags associated with the post (array)
last_comment_time: Unix timestamp of the last comment (integer)
comment_off: Flag indicating if comments are disabled (integer)
liked: Flag indicating if the viewer liked the post (integer)
new_comments: Count of new comments (integer)
canDelete: Flag indicating if the post can be deleted by viewer (integer)
favorite: Flag indicating if the viewer favorited the post (integer)

Data Files

Images referenced in the media and avatar fields are stored in zip archives in corresponding directories (e.g., media images are stored in the media directory) with their original filenames. Each archive contains 1,000 images.

Additional Information

License

This dataset is dedicated to the public domain under the Creative Commons Zero (CC0) license. This means you can:

Use it for any purpose, including commercial projects.
Modify it however you like.
Distribute it without asking permission. No attribution is required, but it's always appreciated!

CC0 license: https://creativecommons.org/publicdomain/zero/1.0/deed.en

To learn more about CC0, visit the Creative Commons website: https://creativecommons.org/publicdomain/zero/1.0/

Dataset Curators

nyuuzyou