metadata

dataset_info:
  features:
    - name: id
      dtype: uint32
    - name: language
      dtype: string
    - name: url
      dtype: string
    - name: title
      dtype: string
    - name: text_markdown
      dtype: string
    - name: text_html
      dtype: string
    - name: author
      dtype: string
    - name: original_author
      dtype: string
    - name: original_url
      dtype: string
    - name: lead_html
      dtype: string
    - name: lead_markdown
      dtype: string
    - name: type
      dtype: string
    - name: time_published
      dtype: uint64
    - name: statistics
      struct:
        - name: commentsCount
          dtype: uint32
        - name: favoritesCount
          dtype: uint32
        - name: readingCount
          dtype: uint32
        - name: score
          dtype: int32
        - name: votesCount
          dtype: int32
        - name: votesCountPlus
          dtype: int32
        - name: votesCountMinus
          dtype: int32
    - name: labels
      sequence: string
    - name: hubs
      sequence: string
    - name: flows
      sequence: string
    - name: tags
      sequence: string
    - name: reading_time
      dtype: uint32
    - name: format
      dtype: string
    - name: complexity
      dtype: string
    - name: comments
      sequence:
        - name: id
          dtype: uint64
        - name: parent_id
          dtype: uint64
        - name: level
          dtype: uint32
        - name: time_published
          dtype: uint64
        - name: score
          dtype: int32
        - name: votes
          dtype: uint32
        - name: message_html
          dtype: string
        - name: message_markdown
          dtype: string
        - name: author
          dtype: string
        - name: children
          sequence: uint64
  splits:
    - name: train
      num_bytes: 19968161329
      num_examples: 302049
  download_size: 3485570346
  dataset_size: 19968161329
task_categories:
  - text-generation
language:
  - ru
  - en
size_categories:
  - 100K<n<1M

Habr dataset

Table of Contents
Description
Usage
Data Instances
Source Data
Personal and Sensitive Information

Description

Summary: Dataset of posts and comments from habr.com.

Script: save_habr.py

Point of Contact: Ilya Gusev

Languages: Russian, English, some programming code.

Usage

Prerequisites:

pip install datasets zstandard jsonlines pysimdjson

Dataset iteration:

from datasets import load_dataset
dataset = load_dataset('IlyaGusev/habr', split="train", streaming=True)
for example in dataset:
    print(example["text_markdown"])

Data Instances

{
  "id": 12730,
  "language": "ru",
  "url": "https://habr.com/ru/post/12730/",
  "text_markdown": "...",
  "text_html": "...",
  "lead_markdown": "...",
  "lead_html": "...",
  "type": "article",
  "labels": [],
  "original_author": null,
  "original_url": null,
  "time_published": 1185962380,
  "author": "...",
  "title": "Хочешь в университет — сделай презентацию",
  "statistics": {
    "commentsCount": 23,
    "favoritesCount": 1,
    "readingCount": 1542,
    "score": 7,
    "votesCount": 15,
    "votesCountPlus": 11,
    "votesCountMinus": 4
  },
  "hubs": [
    "itcompanies"
  ],
  "flows": [
    "popsci"
  ],
  "tags": [
    "PowerPoint",
    "презентация",
    "абитуриенты",
  ],
  "reading_time": 1,
  "format": null,
  "complexity": null,
  "comments": {
    "id": [11653537, 11653541],
    "parent_id": [null, 11653537],
    "level": [0, 1],
    "time_published": [1185963192, 1185967886],
    "score": [-1, 0],
    "votes": [1, 0],
    "message_html": ["...", "..."],
    "author": ["...", "..."],
    "children": [[11653541], []]
  }
}

You can use this little helper to unflatten sequences:

def revert_flattening(records):
    fixed_records = []
    for key, values in records.items():
        if not fixed_records:
            fixed_records = [{} for _ in range(len(values))]
        for i, value in enumerate(values):
            fixed_records[i][key] = value
    return fixed_records

The original JSONL is already unflattened.

Source Data

The data source is the Habr website.
API call example: post 709430.
Processing script is here.

Personal and Sensitive Information

The dataset is not anonymized, so individuals' names can be found in the dataset. Information about the original authors is included in the dataset where possible.