Retracted Articles Dataset Generator

Retracted Articles Dataset Generator is a script focused on finding COVID-19 retracted articles and those who referenced them.

Development

Details about the data search accomplished in 18/04/2021 and later, in 25/07/2021.

Language

Python 3.6

Sources

• Retraction Watch Database
• Semantic Scholar API
• DOI API

Data from each source

• Retraction Watch Database: Title field filtered by “COVID-19” or “coronavirus disease 2019” or “coronavirus 2019” or “SARS- COV-2” or “2019-nCov”. It, then, resulted in 92 articles, from which 9 did not have available DOI, therefore, resulting in 83 from those. Finally, 6 where repeted and had to be excluded, resulting in 77 retracted articles.
• Semantic Scholar API: It was used in order to obtain more information about each retracted article and those who referenced them. From those 77 found in the previous step, 4 were not available in this API, consequently, remaining 73 articles.
• DOI API: Used to obtain the publication date from the articles.

First Dataset

• Format: GraphML
• Total Nodes: 5.884
• Retracted Nodes: 73
• Total Edges: 6.623
• Influential Edges: 423

Second Dataset

• Format: RDF (Turtle)
• Total Articles: 9.070
• Retracted Articles: 131

Ontologies used

• FaBio
• CiTo

Repository Files

• GRAPHML: Containing data from the first data search

  • dataset_generator.py: Python script to create dataset
  • rede-finalv2.gephi: Gephi project with created graph
  • rede-finalv2.graphml: Graph generated by python script
  • img folder: Folder with images generated by Gephi using the created graph

• RDF: Containing data from the second data search

  • dataset_generator.py: Python script to create dataset
  • retracted-2021-07-25.ttl: Turtle file generated by python script
  • retracted.json: Artigos retratados extraídos do Retraction Watch

License

This project uses the MIT license. For more details, read LICENSE.md.