README.md

metadata

license: cc0-1.0
task_categories:
  - feature-extraction
tags:
  - arXiv
size_categories:
  - 1K<n<10K

Manually Annotated arXiv Preprints Dataset for Structured Extraction of Authors and Affiliations

Dataset Description

This dataset contains manually annotated, structured metadata for a random sample of preprints from arXiv. Each entry in the dataset corresponds to a single publication and includes its title, language, arXiv ID, DOI link, a structured list of authors with their respective affiliations, and the corresponding PDF filename.

Data Fields

Each object in the JSON dataset contains the following fields:

• arxiv_id (string): The unique arXiv identifier for the publication (e.g., arXiv:hep-ph/0211259).
• doi (string): The Digital Object Identifier (DOI) link for the publication (e.g., https://doi.org/10.48550/arXiv.hep-ph/0211259).
• title (object): An object containing the title of the publication.
  • text (string): The full title of the paper.
  • lang (string): The language code for the title (e.g., en).
• authors (array): A list of objects, where each object represents an author.
  • name (string): The full name of the author.
  • affiliations (array): A list of strings, where each string is an institutional affiliation for the author.
• filename (string): The filename of the corresponding PDF document, derived from the arXiv ID with forward slashes replaced by underscores (e.g., math_0601203.pdf for arXiv:math/0601203).

Example Data Point

{  
  "arxiv_id": "arXiv:2003.03151",  
  "doi": "https://doi.org/10.48550/arXiv.2003.03151",  
  "title": {  
    "text": "Demographic Bias in Presentation Attack Detection\n\nof Iris Recognition Systems",  
    "lang": "en"  
  },  
  "authors": [  
    {  
      "name": "Naser Damer",  
      "affiliations": [  
        "Fraunhofer Institute for Computer Graphics Research IGD, Darmstadt, Germany",  
        "Mathematical and Applied Visual Computing, TU Darmstadt, Darmstadt, Germanyjo"  
      ]  
    },  
    {  
      "name": "Florian Kirchbuchner",  
      "affiliations": [  
        "Fraunhofer Institute for Computer Graphics Research IGD, Darmstadt, Germany",  
        "Mathematical and Applied Visual Computing, TU Darmstadt, Darmstadt, Germanyjo"  
      ]  
    },  
    {  
      "name": "Arjan Kuijper",  
      "affiliations": [  
        "Fraunhofer Institute for Computer Graphics Research IGD, Darmstadt, Germany",  
        "Mathematical and Applied Visual Computing, TU Darmstadt, Darmstadt, Germanyjo"  
      ]  
    }  
  ],
  "filename": "2003.03151.pdf"
}

Filename Format

The filename field follows a consistent pattern based on the arXiv ID:

• Standard format (YYMM.NNNNN): 2003.03151.pdf
• Category format (category/YYMMNNN): math_0601203.pdf (forward slash replaced with underscore)

Data Splits

This dataset contains predefined train and test splits:

• Training set: 1,400 records (56.18% of total)
  • Contains 400 ArXiv IDs from trained_on_ids.json
  • Contains 1,000 ArXiv IDs from next_up.json
• Test set: 1,092 records (43.82% of total)
  • Contains all remaining ArXiv IDs not in the training set

Source Data

The data was derived from a random sample of 2492 arXiv preprint PDFs. This dataset was manually annotated to indicate the title, title language, author, and affiliation metadata.