Open Science Dataset

Overview

Open Science is a large-scale, permissively licensed text dataset derived from OpenAlex, containing over 100B (105,390,332,599) words words. OpenAlex is an open database of scholarly publications, authors, institutions, and research outputs that serves as a comprehensive source for academic literature.

Key Features

Truly Open: Contains only permissively licensed data suitable for both commercial and non-commercial use
Multilingual Coverage: Predominantly English with significant representation of [additional languages]
Academic Diversity: Encompasses [types of content from OpenAlex]

Dataset Statistics

[Insert key statistics about:]

cc-by: 10,737,025.0 (93.09%) 2025-01-22 16:38:32,175 - INFO - cc-by-sa: 424,556.0 (3.68%) 2025-01-22 16:38:32,175 - INFO - public-domain: 372,583.0 (3.23%)

Total number of documents
Word count distribution
Language distribution
License type
Temporal coverage
[Other relevant metrics]

Dataset Structure

Data Fields

Field	Type	Description
[field1]	[type]	[description]
[field2]	[type]	[description]
...	...	...

How to Use

Installation

from datasets import load_dataset
dataset = load_dataset('PleIAs/openscource')

Data Filtering Capabilities

Language: Filter by specific languages using metadata
Temporal: Select data based on publication dates

Considerations and Limitations

While permissively licensed, users should verify compatibility with their specific use case
Language distribution may not be uniform
Historical texts may contain dated terminology or perspectives
[Other important considerations]

Acknowledgements

This dataset was developed with support from: (Update this, this is from common corpus)

AI Alliance
Jean Zay (Eviden, Idris)
Nvidia Inception program
Nebius AI
Tracto AI
Mozilla Special recognition to:
LANGU:IA (French Ministry of Culture and DINUM state startup)
Alliance for Language technologies EDIC (ALT-EDIC)
Wikimedia Enterprise
Open science LLM community (Eleuther AI, Allen AI, HuggingFace)