Mining tortured acronyms from the scientific literature

Alexandre Clausse (1) , Guillaume Cabanac (1, 2, 3) , Pascal Cuxac (4) , Cyril Labbé (5, 6)

1 IRIT-IRIS - Recherche d’Information et Synthèse d’Information
2 IUF - Institut universitaire de France
3 UT3 - Université Toulouse III - Paul Sabatier
4 INIST - Institut de l'information scientifique et technique
5 SIGMA - Systèmes d’Information - inGénierie et Modélisation Adaptables
6 UGA - Université Grenoble Alpes

The 'Problematic Paper Screener' (PPS, WCRI'22, https://doi.org/10.48550/arXiv.2210.04895) supports the human re-assessment of scientific articles flagged as suspicious. The 'tortured detector' tabulates 12k papers containing tortured phrases: established scientific concepts paraphrased with synonyms, such as 'butt-centric waterway' for 'anal canal.' Some acronyms are even tortured, such as 'Convolutional Brain Organisation (CNN)' for 'Convolutional Neural Network (CNN).' This abstract tackles the following task: discover and classify all acronyms from any given article: tortured or genuine.

Keywords: Tortured phrases; Acronyms

Domains: Digital Libraries [cs.DL]

Alexandre Clausse, Guillaume Cabanac, Pascal Cuxac, Cyril Labbé. Mining tortured acronyms from the scientific literature. 2023. ⟨hal-04311600⟩