Episciences Adopts Problematic Paper Screener as a Tool for Verifying Bibliographic References

Written by Agnès Magron

Have you heard of the Problematic Paper Screener (PPS) ?  The PPS is a tool designed to identify and filter out potentially problematic scientific articles, such as those containing methodological errors, plagiarism, data manipulation, or conflicts of interest.

Developed by G. Cabanac, C. Labbé, and A. Magazinov , it provides a database of over one million references identified as containing problematic elements, categorised according to a typology that distinguishes between anomalies of varying severity – ranging from articles written entirely by AI to articles that are valid but contain references with anomalies (“Feet of clay”).

This database is now available among the Episciences tools. It is used to analyze citations extracted from articles, detect and then display problematic citations, either for correction or for informational purposes.

Verifying references: An ethical issue
Verifying the accuracy and validity of the bibliographic references cited in a scientific article is an important step in the editorial process. The reliability and integrity of the scientific work presented in the article can be seriously called into question by references pointing to retracted works, or works containing obvious errors, manipulation, or even fraud. Episciences is committed to providing tools that facilitate the work of journals and guide them toward editorial practices that adhere to the COPE ethical principles to which it subscribes.
Toward a more comprehensive and automated verification solution

The management of bibliographic references was already possible using the Episciences Citations tool. This module, which has recently been updated and modernised, is available to authors and journal editorial teams for processing references from previously published articles. It enables the semi-automated management of these bibliographic references by extracting them using Grobid, or importing them via a BibTeX file, a list of Crossref DOIs, or through the Semantic Scholar API. They can then be manually validated and corrected before being made available to readers of the articles.

The platform takes this process a step further by incorporating automatic reference verification right after the submission is received, to help assess its quality, and subsequently by automatically updating any references that are retracted or change status after publication. This automatic check is performed using the PPS database. It allows editorial teams, authors, and reviewers to view problematic references on the article management page and in the Episciences Citations application.

The ‘Episciences Citations’ interface. It has two distinct sections: on the left, the PDF in preview mode ; on the right, the references identified and extracted automatically by Grobid.

‘Episciences Citations’ interface

Thus, this new development enhances the effectiveness of Episciences Citations by leveraging an extensive and reliable database to verify the quality of references, without altering the journals’ teams’ usual work habits.

Work is currently underway to cover all publications on the platform

The module is already in place to analyze new submissions from all journals. For the more than 8,000 manuscripts that have already been published or are in the process of being published, the first step involved analyzing nearly 170,000 references from an initial data extraction. The results of this analysis identified 226 references indexed in the PPS, the majority of which fall into the “Feet of clay” category. The remaining citations will be extracted and analyzed in the coming months.

Verifying the reliability of citations is no longer an option but a necessity to ensure scientific integrity. With Problematic Paper Screener, Episciences offers an effective solution: it saves time, proactively detects problematic articles, and contributes to building trust in research.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.