---
language:
- en
dataset_info:
  features:
  - name: id
    dtype: string
  - name: submitter
    dtype: string
  - name: authors
    dtype: string
  - name: title
    dtype: string
  - name: comments
    dtype: string
  - name: journal-ref
    dtype: string
  - name: doi
    dtype: string
  - name: report-no
    dtype: string
  - name: categories
    dtype: string
  - name: license
    dtype: string
  - name: abstract
    dtype: string
  - name: versions
    list:
    - name: version
      dtype: string
    - name: created
      dtype: string
  - name: update_date
    dtype: timestamp[s]
  - name: authors_parsed
    sequence:
      sequence: string
  splits:
  - name: train
    num_bytes: 3538777556
    num_examples: 2367176
  download_size: 1992564422
  dataset_size: 3538777556
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
license: cc0-1.0
size_categories:
- 1M<n<10M
---

# ArXiv Dataset
## Overview

## Dataset Description

- **Homepage:** [Kaggle arXiv Dataset Homepage](https://www.kaggle.com/Cornell-University/arxiv)
- **Repository:** 
- **Paper:** [On the Use of ArXiv as a Dataset](https://arxiv.org/abs/1905.00075)

This dataset is a comprehensive collection of metadata from the ArXiv repository, a widely-recognized open-access archive offering access to scholarly articles in various fields of science. It covers a broad range of subjects from physics and computer science to mathematics, statistics, electrical engineering, quantitative biology, and economics.

The dataset hosted here is derived from the original ArXiv dataset available on Kaggle, which includes metadata for approximately 2.2 million articles. The metadata encompasses various features such as article titles, authors, categories, abstracts, and full text in PDF format.

This rich repository of scholarly articles provides a valuable resource for data analysis, trend identification, and development of machine learning models. It can facilitate applications like trend analysis, paper recommendation systems, category prediction, co-citation network analysis, knowledge graph construction, and semantic search interfaces.

The data is particularly suited for those interested in natural language processing and text analytics within the academic domain.

## Dataset Composition
The dataset is divided into multiple .parquet files, structured to enable efficient access and analysis. Each file contains a subset of the entire dataset, allowing users to work with manageable portions of data as needed.

## Original Dataset Source
The dataset is based on the **[ArXiv dataset hosted on Kaggle](https://www.kaggle.com/datasets/Cornell-University/arxiv)**, provided by Cornell University. It represents a snapshot of the ArXiv metadata.


### License
This dataset is made available under the CC0: Public Domain License. The original dataset from ArXiv, as provided by Cornell University on Kaggle, is also under the same license, allowing for unrestricted use and distribution.

### Citation and Acknowledgments
When using or citing this dataset, please acknowledge the original source of the data: ArXiv dataset on Kaggle, maintained and operated by Cornell University.