HenryQW's picture
chore: update README and license
898c33d unverified
metadata
license: cc0-1.0
dataset_info:
  features:
    - name: handle
      dtype: string
    - name: body
      dtype: string
    - name: media
      list:
        - name: address
          dtype: string
        - name: mime_type
          dtype: string
    - name: profile_id
      dtype: string
    - name: publication_id
      dtype: string
    - name: timestamp
      dtype: string
  splits:
    - name: batch_0
      num_bytes: 220122462
      num_examples: 891118
    - name: batch_1
      num_bytes: 885209133
      num_examples: 3000000
    - name: batch_2
      num_bytes: 640808994
      num_examples: 2172208
    - name: batch_3
      num_bytes: 924120947
      num_examples: 3000000
    - name: batch_4
      num_bytes: 762173887
      num_examples: 2396247
    - name: batch_5
      num_bytes: 7506223
      num_examples: 22735
    - name: batch_6
      num_bytes: 335272
      num_examples: 842
    - name: batch_7
      num_bytes: 8748603
      num_examples: 18352
  download_size: 2222667988
  dataset_size: 3449025521
configs:
  - config_name: default
    data_files:
      - split: batch_0
        path: data/batch_0-*
      - split: batch_1
        path: data/batch_1-*
      - split: batch_2
        path: data/batch_2-*
      - split: batch_3
        path: data/batch_3-*
      - split: batch_4
        path: data/batch_4-*
      - split: batch_5
        path: data/batch_5-*
      - split: batch_6
        path: data/batch_6-*
      - split: batch_7
        path: data/batch_7-*

A High Quality Open Web Content Dataset

This dataset is curated by the RSS3 Network. It contains a large collection of content from a variety of decentralized platforms such as Farcaster and Lens.

The dataset has been structured and indexed by RSS3 Nodes, and is provided in a structured format for easy access and analysis.

All content is available on the RSS3 Network's Data Sublayer.

RSS3 ecosystem projects have been leveraging the dataset to build various applications and services, including finetuning large language models, training recommendation systems, and more.

Fields

  1. handle: The handle of the author on the corresponding platform.
  2. body: The text content of the post.
  3. media: A list of media objects associated with the post.
    • address: The URL of the media.
    • mime_type: The MIME type of the media.
  4. profile_id: The profile id of the author on the corresponding platform.
  5. publication_id: The publication id of the post on the corresponding platform.
  6. timestamp: The timestamp of the post on the corresponding platform. When not available, it is the time in which the data was indexed and structured by RSS3 Nodes.

Alternative Access

For enhanced filtering options and more control over data granularity, you may access the dataset through the RSS3 Network. For more details, please refer to our documentation: https://docs.rss3.io.

Other Resources

In addition to providing data, RSS3 offers SDKs and frameworks for building AI solutions, learn more: https://docs.rss3.io/.

As a proud member of the Open Information Initiative, RSS3 also sponsors the Open Information Grant. If you have a great idea, apply here: https://openinformation.io/grant.

License and Disclaimer

This dataset is marked with CC0 1.0.

By marking the work with a CC0 public domain dedication, the creator is giving up their copyright and allowing reusers to distribute, remix, adapt, and build upon the material in any medium or format, even for commercial purposes.

RSS3 Network and its Node operators does not own, produce and/or endorse any of the content in this dataset. All content is owned by the respective authors and/or platforms, which is provided in a decentralized manner with unrestricted access. At the time of publication, the respective platforms did not restrict the redistribution of the content via their terms of service and/or licenses.