README.md

metadata

annotations_creators:
  - no-annotation
language:
  - en
language_creators:
  - found
license: mit
multilinguality:
  - monolingual
pretty_name: Medium Articles Dataset
size_categories:
  - n>1K
source_datasets:
  - original
tags:
  - medium
  - articles
  - blog-posts
task_categories:
  - text-classification
  - text-generation
task_ids:
  - topic-classification
  - language-modeling
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
dataset_info:
  features:
    - name: audioVersionDurationSec
      dtype: float64
    - name: codeBlock
      dtype: string
    - name: codeBlockCount
      dtype: float64
    - name: collectionId
      dtype: string
    - name: createdDate
      dtype: string
    - name: createdDatetime
      dtype: string
    - name: firstPublishedDate
      dtype: string
    - name: firstPublishedDatetime
      dtype: string
    - name: imageCount
      dtype: float64
    - name: isSubscriptionLocked
      dtype: bool
    - name: language
      dtype: string
    - name: latestPublishedDate
      dtype: string
    - name: latestPublishedDatetime
      dtype: string
    - name: linksCount
      dtype: float64
    - name: postId
      dtype: string
    - name: readingTime
      dtype: float64
    - name: recommends
      dtype: float64
    - name: responsesCreatedCount
      dtype: float64
    - name: socialRecommendsCount
      dtype: float64
    - name: subTitle
      dtype: string
    - name: tagsCount
      dtype: float64
    - name: text
      dtype: string
    - name: title
      dtype: string
    - name: totalClapCount
      dtype: float64
    - name: uniqueSlug
      dtype: string
    - name: updatedDate
      dtype: string
    - name: updatedDatetime
      dtype: string
    - name: url
      dtype: string
    - name: vote
      dtype: bool
    - name: wordCount
      dtype: float64
    - name: publicationdescription
      dtype: string
    - name: publicationdomain
      dtype: string
    - name: publicationfacebookPageName
      dtype: string
    - name: publicationfollowerCount
      dtype: float64
    - name: publicationname
      dtype: string
    - name: publicationpublicEmail
      dtype: string
    - name: publicationslug
      dtype: string
    - name: publicationtags
      dtype: string
    - name: publicationtwitterUsername
      dtype: string
    - name: tag_name
      dtype: string
    - name: slug
      dtype: string
    - name: name
      dtype: string
    - name: postCount
      dtype: float64
    - name: author
      dtype: string
    - name: bio
      dtype: string
    - name: userId
      dtype: string
    - name: userName
      dtype: string
    - name: usersFollowedByCount
      dtype: float64
    - name: usersFollowedCount
      dtype: float64
    - name: scrappedDate
      dtype: float64
    - name: claps
      dtype: string
    - name: reading_time
      dtype: float64
    - name: link
      dtype: string
    - name: authors
      dtype: string
    - name: timestamp
      dtype: string
    - name: tags
      dtype: string
  splits:
    - name: train
      num_bytes: 2654611084
      num_examples: 444593
  download_size: 1482558340
  dataset_size: 2654611084

Medium Articles Dataset Generator

This project combines multiple datasets from Kaggle and Hugging Face to create a comprehensive collection of Medium articles. The combined dataset is available on Hugging Face Hub.

Dataset Description

This dataset is a unique compilation that not only combines multiple sources but also ensures data quality through normalization and deduplication. A key feature is that all entries in the text column are unique - there are no duplicate articles in the final dataset.

Data Sources:

Kaggle Sources:

• aiswaryaramachandran/medium-articles-with-content
• hsankesara/medium-articles
• meruvulikith/1300-towards-datascience-medium-articles-dataset

Hugging Face Sources:

• fabiochiu/medium-articles
• Falah/medium_articles_posts

Features

• Combines multiple data sources into a single, unified dataset
• Ensures uniqueness: Each article appears only once in the dataset
• Quality control:
  • Removes duplicate entries based on article text
  • Handles missing values
  • Normalizes data format
• Saves the final dataset in efficient Parquet format
• Publishes the dataset to Hugging Face Hub

Requirements

pip install datasets
pip install kagglehub huggingface_hub tqdm

Usage

1. Set up your Hugging Face authentication token
2. Run the script:

python combined_medium_ds_generator.py

Data Processing Steps

1. Downloads datasets from Kaggle and Hugging Face
2. Normalizes each dataset by:
   • Removing null values
   • Eliminating duplicates
   • Standardizing column names
3. Combines all datasets into a single DataFrame
4. Saves the result as a Parquet file
5. Uploads the final dataset to Hugging Face Hub

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Author

• @Alaamer

Acknowledgments

Special thanks to the original dataset creators:

• aiswaryaramachandran
• hsankesara
• meruvulikith
• fabiochiu
• Falah