The DOI matches its Crossref record; 72 of the 72 references it cites were found in a registry or an index; 7 of the 23 sources its claims cite were checked for support.
Can ChatGPT be used to predict citation counts, readership, and social media interaction? An exploration among 2222 scientific abstracts
Open accessAbstract
This study explores the potential of ChatGPT, a large language model, in scientometrics by assessing its ability to predict citation counts, Mendeley readers, and social media engagement. In this study, 2222 abstracts from PLOS ONE articles published during the initial months of 2022 were analyzed using ChatGPT-4, which used a set of 60 criteria to assess each abstract. Using a principal component analysis, three components were identified: Quality and Reliability, Accessibility and Understandability, and Novelty and Engagement. The Accessibility and Understandability of the abstracts correlated with higher Mendeley readership, while Novelty and Engagement and Accessibility and Understandability were linked to citation counts (Dimensions, Scopus, Google Scholar) and social media attention. Quality and Reliability showed minimal correlation with citation and altmetrics outcomes. Finally, it was found that the predictive correlations of ChatGPT-based assessments surpassed traditional readability metrics. The findings highlight the potential of large language models in scientometrics and possibly pave the way for AI-assisted peer review.
Overview
CoChat Summary
TL;DR
This study used ChatGPT-4 to rate 2222 PLOS ONE abstracts on 60 criteria and tested whether the resulting components predict citations, Mendeley readers, and social media attention. It found weak-to-moderate correlations, with ChatGPT-based assessments outperforming traditional readability metrics.
Key findings
- Principal component analysis identified three components: Quality and Reliability, Accessibility and Understandability, and Novelty and Engagement; the 30 items and their 30 antonyms showed inverse loadings correlated at r = −0.91, −0.92, and −0.93.
- Accessibility and Understandability correlated with Mendeley readers (ρ = 0.40), Twitter mentions (ρ = 0.18), news mentions (ρ = 0.10), blog mentions (ρ = 0.07), Reddit mentions (ρ = 0.06), and citations (ρ = 0.08 to 0.17).
- Novelty and Engagement correlated with citations (ρ = 0.18), Twitter mentions (ρ = 0.23), blog mentions (ρ = 0.15), Reddit mentions (ρ = 0.14), Mendeley readers (ρ = 0.12), and news mentions (ρ = 0.09).
- Quality and Reliability showed weak correlations with citations (ρ = 0.05 to 0.10) and altmetrics; traditional readability indexes had a strongest Spearman correlation of only 0.13, while text attributes predicted Mendeley readers up to ρ = 0.29.
- ChatGPT-4 item means were highly consistent across runs (r > 0.998), but individual-abstract reliability averaged r = 0.27, ranging from r = 0.00 for “Ethical” to r = 0.57 for “Easy to understand.”
Methods
- 2222 PLOS ONE article abstracts published January–February 2022 were downloaded via Scopus; citation and altmetric data were extracted on January 4, 2024 from Scopus, Altmetric, Dimensions, and Google Scholar via Publish or Perish, with missing Mendeley reader counts manually inserted.
- ChatGPT-4 API rated each abstract on 60 items (30 items plus 30 antonyms) from 0 to 100, with temperature set to 0 and items presented in random order; four runs were averaged, then PCA with Varimax rotation retained three components.
- Spearman rank-order correlations tested component scores against abstract length, blog/news/Twitter/Reddit mentions, Mendeley readers, and Dimensions/Scopus/Google Scholar citations; classical readability indexes were computed for comparison.
Limitations
- Stated: an abstract may not represent the entire paper, and it is unclear whether the abstract is causal or an epiphenomenon reflecting overall paper quality; future work could analyze full texts and titles.
- Stated: the dataset of 2222 abstracts was smaller than desirable, cost about 0.04 USD per abstract and roughly 27 h / USD 91 total, and the papers had only 22–24 months to accumulate citations and altmetrics.
- Stated: individual-abstract ChatGPT-4 scores had low reliability (average r = 0.27) and were sensitive to item order; some ratings appeared superficial. Evident from the text: the sample is limited to PLOS ONE articles from early 2022, and the reported predictive correlations are only weak to moderate.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 1 Not valid · 10 Not checked · 54 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“Due to the potential presence of outliers (e.g., a small number of articles attracting a large number of tweets), Spearman rank-order correlation coefficients were used for determining associations with citations and altmetrics. The Spearman rank-order correlation is robust against the influence of outliers”
What the sources say
- Supportedcites Croux & Dehon, 2010 source
The Introduction states that the influence function of Spearman correlation is bounded and says this confirms its robustness to outliers. That supports the cited rationale for choosing Spearman correlation, without making a claim about the citing study’s particular data. (LLM, conf=0.94)
- Supportedcites De Winter et al., 2016 source
The section “r p Versus r s When There are Outliers” states that Spearman’s correlation is robust to outliers because ranking limits the effect of extreme values. The Discussion likewise says the simulations found Spearman’s correlation robust while Pearson’s was sensitive to an outlier. Page numbers were not available in our copy. (LLM, conf=0.95)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“the predictive correlations of ChatGPT-based assessments surpassed traditional readability metrics.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This component also displayed a moderate positive correlation with the number of citations (ρ = 0.18).”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“abstracts that achieved higher scores in Novelty and Engagement received lower readability scores (ρ up to 0.28).”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“abstracts that received higher scores on the Accessibility and Understandability component were associated with higher readability scores (ρ up to -0.21)”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
29 more claims
Create a free account to see all 69 claims and to verify the 9 references still open.
- Probably not a substantial claim. 29
- It may state two separate claims as one. 15
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Discussion · p. 14
“with Spearman's ρ values ranging between 0.08 and 0.18.”
§ Predicting altmetrics scores and citation counts from classical readability scores · p. 10
“| Number of Google Scholar citations | 8.65 | 11.94 | 0.05 | 0.17 | 0.18 |”
Citations
Cites
Statistical Power Analysis for the Behavioral Sciences
Muller, K. (1989) · Technometrics · doi:10.1080/00401706.1989.10488618 · Cited by 99,167
The Scree Test For The Number Of Factors
Cattell, R. B. (1966) · Multivariate Behavioral Research · doi:10.1207/s15327906mbr0102_10 · Cited by 13,554 · cited 1×
“The number of components to retain was based on a scree plot (Cattell, 1966).”
Flesch, R. (1948) · Journal of Applied Psychology · doi:10.1037/h0057532 · Cited by 5,391 · cited 2×
“These features served as building blocks for calculating various classical readability indexes, including the Automated Readability Index (Senter & Smith, 1967), SMOG grade (McLaughlin, 1969), Flesch-Kincaid grade level (Kincaid et al., 1975), Flesch Reading Ease (Flesch, 1948), Coleman-Liau index (Coleman & Liau, 197…”
Kincaid, J. P., Fishburne, Jr., Robert P., R., Richard L., C., & Brad, S. (1975) · doi:10.21236/ada006655 · Cited by 3,497 · cited 1×
“…classical readability indexes, including the Automated Readability Index (Senter & Smith, 1967), SMOG grade (McLaughlin, 1969), Flesch-Kincaid grade level (Kincaid et al., 1975), Flesch Reading Ease (Flesch, 1948), Coleman-Liau index (Coleman & Liau, 1975), and Gunning fog index (Gunning, 1952).”
An effect size primer: A guide for clinicians and researchers.
Ferguson, C. J. (2009) · Professional Psychology: Research and Practice · doi:10.1037/a0015808 · Cited by 3,467 · cited 1×
“Later, Ferguson (2009) recommended a minimum correlation of r = 0.20 to deem an effect practically significant.”
InfluentialEffect size guidelines for individual differences researchers
Gignac, G. E., & Szodorai, E. T. (2016) · Personality and Individual Differences · doi:10.1016/j.paid.2016.06.069 · Cited by 2,860 · cited 1×
“On the other hand, a meta-analysis by Gignac and Szodorai (2016) showed that correlations of 0.50 or stronger are rare in individual-differences research, and proposed a more realistic guideline of r = 0.10, r = 0.20, and r = 0.30 representing small, medium, and large effects.”
InfluentialSMOG Grading - A New Readability Formula.
McLaughlin, G. H. (1969) · The Journal of Reading · Cited by 2,081
The Technique of Clear Writing.
Gunning, R. C. (1968) · CERN Document Server (European Organization for Nuclear Research) · Cited by 1,641
Sparks of Artificial General Intelligence: Early experiments with GPT-4
Bubeck, S., Chandrasekaran, V., Eldan, R., Gehrke, J., Horvitz, E., Kamar, E., Lee, P., Lee, Y. T., Li, Y., Lundberg, S., et al. (2023) · arXiv · doi:10.48550/arxiv.2303.12712 · Cited by 1,526 · cited 1×
“The concern of contamination in benchmark analyses has also been recognized by other researchers (Aiyappa et al., 2023; Bubeck et al., 2023; De Winter, 2023; OpenAI, 2023a).”
Citations, Citation Indicators, and Research Quality: An Overview of Basic Concepts and Theories
Aksnes, D. W., Langfeldt, L., & Wouters, P. (2019) · Sage Open · doi:10.1177/2158244019829575 · Cited by 1,257 · cited 1×
“Citations are often used as an indicator of the impact of research contributions, allowing researchers, institutions, funding agencies, and policymakers to make informed decisions (e.g., Aksnes et al., 2019; Caon et al., 2020).”
Cited by
Thelwall, M. A. (2024) · Journal of Data and Information Science · doi:10.2478/jdis-2025-0011 · Cited by 70
The use of ChatGPT for personality research: Administering questionnaires using generated personas
de Winter, J., Driessen, T., & Dodou, D. (2024) · Personality and Individual Differences · doi:10.1016/j.paid.2024.112729 · Cited by 65
Self-citationChatGPT in Education: Empowering Educators through Methods for Recognition and Assessment
de Winter, J., Dodou, D., & Stienen, A. H. A. (2023) · Informatics · doi:10.3390/informatics10040087 · Cited by 49
“To resolve this, De Winter [33] used a self-consistency prompting method, where the items were repeatedly presented in a different order and the scores were averaged.”
BackgroundMethodInfluentialSelf-citationThelwall, M. (2025) · Scientometrics · doi:10.1007/s11192-025-05361-8 · Cited by 33
“For the related goal of novelty scoring, LLMs have been shown to have some ability to predict novelty scores for computer science papers from one conference from the introduction, results, and discussion but not from the full text (Wu et al., 2025; see also: de Winter, 2024).”
Goyanes, M., Lopezosa, C., & Piñeiro-Naval, V. (2025) · Scientometrics · doi:10.1007/s11192-025-05377-0 · Cited by 29
“As such, AI technologies are emerging as essential tools, significantly accelerating the pace of scientific research and improving, if used well, the overall efficiency of scholars (Bornmann et al., 2024; de Winter, 2024; Lehr et al., 2024; Zhang et al., 2024).”
Metadata
- Authors
- de Winter, J.
- ORCID
- de Winter, Joost
- Venue
- Scientometrics
- Published
- 2024-02-15
- Volume
- 129
- Issue
- 4
- Pages
- 2469-2487
- ISSN
- 0138-9130 (ISSN-L)
- 1588-2861
- OpenAlex
- W4391839308
- Semantic Scholar
- e3048c7eb611680a66728568af8b1ca75e20c006
- Cited by
- 34
- References
- 63
- Citations by year
- 2026: 14 · 2025: 13 · 2024: 6
- FWCI
- 9.25
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Scientometrics (publishedVersion, cc-by)
- RePEc: Research Papers in Economics (submittedVersion)
- Research Repository (Delft University of Technology) (submittedVersion, cc-by)
- Topics
- Social Sciences › Library and Information Sciences › scientometrics and bibliometrics research
- Physical Sciences › Computer Science › Information Systems › Web visibility and informetrics
- Social Sciences › Library and Information Sciences › Academic Writing and Publishing
- Keywords
- ChatGPT, large language models, citation prediction, Mendeley readership, social media engagement, altmetrics, scientific abstracts, readability, AI-assisted peer review, PLOS ONE
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Can ChatGPT be used to predict citation counts, readership, and social media interaction? An exploration among 2222 scientific abstracts. https://app.cochat.ai/papers/10.1007/s11192-024-04939-y
- The paper
- de Winter, J. (2024). Can ChatGPT be used to predict citation counts, readership, and social media interaction? An exploration among 2222 scientific abstracts. Scientometrics. https://doi.org/10.1007/s11192-024-04939-y
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.