Prevalence of nonsensical algorithmically generated papers in the scientific literature
Open accessThe DOI matches its Crossref record; 7 of the 7 references it cites were found in a registry or an index.
Abstract
In 2014 leading publishers withdrew more than 120 nonsensical publications automatically generated with the SCIgen program. Casual observations suggested that similar problematic papers are still published and sold, without follow‐up retractions. No systematic screening has been performed and the prevalence of such nonsensical publications in the scientific literature is unknown. Our contribution is 2‐fold. First, we designed a detector that combs the scientific literature for grammar‐based computer‐generated papers. Applied to SCIgen, it has a 83.6% precision. Second, we performed a scientometric study of the 243 detected SCIgen‐papers from 19 publishers. We estimate the prevalence of SCIgen‐papers to be 75 per million papers in Information and Computing Sciences. Only 19% of the 243 problematic papers were dealt with: formal retraction (12) or silent removal (34). Publishers still serve and sometimes sell the remaining 197 papers without any caveat. We found evidence of citation manipulation via edited SCIgen bibliographies. This work reveals metric gaming up to the point of absurdity: fraudsters publish nonsensical algorithmically generated papers featuring genuine references. It stresses the need to screen papers for nonsense before peer‐review and chase citation manipulation in published papers. Overall, this is yet another illustration of the harmful effects of the pressure to publish or perish.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Martín-Martín, A., Thelwall, M., Orduna-Malea, E., & Delgado López-Cózar, E. (2021) · Scientometrics · doi:10.1007/s11192-020-03690-4 · Cited by 1,018
Visser, M., van Eck, N. J., & Waltman, L. (2021) · Quantitative Science Studies · doi:10.1162/qss_a_00112 · Cited by 865
The Google scholar experiment: How to index false papers and manipulate bibliometric indicators
Delgado López‐Cózar, E., Robinson‐García, N., & Torres‐Salinas, D. (2014) · Journal of the Association for Information Science and Technology · doi:10.1002/asi.23056 · Cited by 272
Dimensions: Bringing down barriers between scientometricians and data
Herzog, C., Hook, D., & Konkiel, S. (2020) · Quantitative Science Studies · doi:10.1162/qss_a_00020 · Cited by 219 · cited 12×
“We used the DOIs included in the Dimensions results to get to the landing pages of the publishers.”
BackgroundMethodInfluentialHarzing, A.-W. (2019) · Scientometrics · doi:10.1007/s11192-019-03114-y · Cited by 164
Machine Learning Methods for Stylometry: Authorship Attribution and Author Profiling
Savoy, J. (2020) · doi:10.1007/978-3-030-53360-1 · Cited by 67 · cited 1×
“They relate to authorship attribution and profiling tasks, and we refer to (Savoy, 2020) for a comprehensive review on that topic.”
BackgroundPubPeer: Scientific Assessment Without Metrics
Barbour, B., & Stell, B. M. (2020) · Gaming the Metrics · doi:10.7551/mitpress/11087.003.0015 · Cited by 22
Research works printed but not linked: 30
- 10.1007/s11192-015-1637-z · DOI · Paper page
- 10.7551/mitpress/11087.003.0018 · DOI · Paper page
- Avros R. &Volkovich Z.(2018).Detection of computer‐generated papers using oneclass SVM and cluster approaches. InMLDM'18: Proceedings of the International Conference on Machine Learning and Data Mining in Pattern Recognition(Vol.LNCS 10935 pp.42–55).https://doi.org/10.1007/978-3-319-96133-0_4 · doi:10.1007/978-3-319-96133-0_4
- 10.1038/nature03653 · DOI
- 10.1126/science.348.6230.18 · DOI · Paper page
- 10.1002/asi.23445 · DOI · Paper page
- 10.1126/science.aal0625 · doi:10.1126/science.aal0625
- 10.12688/f1000research.23847.2 · DOI · Paper page
- Dalkilic M. M. Clark W. T. Costello J. C. &Radivojac P.(2006).Using compression to identify classes of inauthentic texts. InProceedings of the 2006 SIAM International Conference on Data Mining.https://doi.org/10.1137/1.9781611972764.69 · DOI
- 10.1007/s11948-014-9521-4 · DOI
- 10.1038/d41586-018-04190-5 · DOI
- 10.1038/508044a · DOI
- 10.3233/isu-160799 · doi:10.3233/isu-160799
- 10.1089/blr.2009.9918 · doi:10.1089/blr.2009.9918
- Labbé C., 2010, Ike Antkare, one of the great stars in the scientific firmament, ISSI Newsletter, 6, 48
- 10.1007/s11192-012-0781-y · DOI · Paper page
- 10.1007/978-3-319-24403-7_8 · DOI
- Lavoie A. &Krishnamoorthy M.(2010 August 4).Algorithmic detection of computer generated text. arXiv. Retrieved fromhttps://arxiv.org/abs/1008.0706
- 10.1109/tip.1996.8100841 · doi:10.1109/tip.1996.8100841
- 10.1038/s42256-020-0223-0 · DOI
- Nguyen M. T.(2018).Detection of automatically generated texts(doctoral dissertation Université Grenoble Alpes). Retrieved fromhttps://tel.archives-ouvertes.fr/tel-01919207
- 10.1007/s11192-018-2789-4 · DOI
- 10.1038/d41586-019-02142-1 · DOI
- Schneegans S., 2015, UNESCO science report: Towards 2030 (Tech. Rep.)
- 10.1038/nature.2014.16269 · DOI
- 10.1038/nature.2014.14763 · DOI · Paper page
- 10.1007/978-3-319-25087-8_32 · DOI
- 10.1109/KESE.2009.62 · DOI
- 10.1145/1044956.1044958 · DOI
- 10.1145/1255175.1255220 · DOI
Cited by
An exploratory survey about using ChatGPT in education, healthcare, and research
Hosseini, M., Gao, C. A., Liebovitz, D. M., Carvalho, A. M., Ahmad, F. S., Luo, Y., MacDonald, N., Holmes, K. L., & Kho, A. (2023) · doi:10.1101/2023.03.31.23287979 · Cited by 238
An exploratory survey about using ChatGPT in education, healthcare, and research
Hosseini, M., Gao, C. A., Liebovitz, D. M., Carvalho, A. M., Ahmad, F. S., Luo, Y., MacDonald, N., Holmes, K. L., & Kho, A. (2023) · PLOS ONE · doi:10.1371/journal.pone.0292216 · Cited by 238
Machine Generated Text: A Comprehensive Survey of Threat Models and Detection Methods
Crothers, E., Japkowicz, N., & Viktor, H. (2022) · arXiv · doi:10.48550/arxiv.2210.07321 · Cited by 196
“Recall from § III-D1 that machine-generated scientific papers have been well-documented since the release of SCIgen in 2005 [7], [134], [135].”
InfluentialMachine-Generated Text: A Comprehensive Survey of Threat Models and Detection Methods
Crothers, E. N., Japkowicz, N., & Viktor, H. L. (2023) · IEEE Access · doi:10.1109/access.2023.3294090 · Cited by 196
“Recall from § III-D1 that machine-generated scientific papers have been well-documented since the release of SCIgen in 2005 [7], [134], [135].”
InfluentialCabanac, G., Labbé, C., & Magazinov, A. (2021) · arXiv · doi:10.48550/arxiv.2107.06751 · Cited by 224
Self-citation
Metadata
- Authors
- Cabanac, G., & Labbé, C.
- Venue
- Journal of the Association for Information Science and Technology
- Publisher
- Wiley
- Published
- 2021-05-25
- Volume
- 72
- Issue
- 12
- Pages
- 1461-1476
- ISSN
- 2330-1635 (ISSN-L)
- 2330-1643
- OpenAlex
- W3165883805
- Cited by
- 83
- References
- 7
- FWCI
- 17.61
- Type
- journal article
- Language
- en
- Open access
- hybrid
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.