Daniel Acuna
@danielacuna.bsky.social
Assoc. Prof. Computer Science at the University of Colorado, Boulder. Prev iSchool, Syracuse University. Postdoc Northwestern University and Ability Lab. PhD Computer Science UMN, Twin Cities
Applied to journals that haven't been vetted before , we predict more than 1000 new potentially questionable journals.
They have collectively published over 500K articles, cited millions of times, and acknowledged major funders in US, China, and Japan
They have collectively published over 500K articles, cited millions of times, and acknowledged major funders in US, China, and Japan
August 27, 2025 at 6:31 PM
Applied to journals that haven't been vetted before , we predict more than 1000 new potentially questionable journals.
They have collectively published over 500K articles, cited millions of times, and acknowledged major funders in US, China, and Japan
They have collectively published over 500K articles, cited millions of times, and acknowledged major funders in US, China, and Japan
We tried several methods, including simple regression, random forest and deep learning.
- Bibliometric features alone: PRC AUC ≈0.64
- Combined model: PRC AUC ≈0.79
Agreement with DOAJ guidelines and expert reviewers was strong, though false positives remain.
- Bibliometric features alone: PRC AUC ≈0.64
- Combined model: PRC AUC ≈0.79
Agreement with DOAJ guidelines and expert reviewers was strong, though false positives remain.
August 27, 2025 at 6:31 PM
We tried several methods, including simple regression, random forest and deep learning.
- Bibliometric features alone: PRC AUC ≈0.64
- Combined model: PRC AUC ≈0.79
Agreement with DOAJ guidelines and expert reviewers was strong, though false positives remain.
- Bibliometric features alone: PRC AUC ≈0.64
- Combined model: PRC AUC ≈0.79
Agreement with DOAJ guidelines and expert reviewers was strong, though false positives remain.
We trained models on ~15,000 journals labeled by DOAJ (12,869 legitimate vs 2,536 removed). Features included:
- Website content (editorial boards, policies)
- Website design (HTML structure, screenshots)
- Bibliometrics (citations, author metrics)
- Website content (editorial boards, policies)
- Website design (HTML structure, screenshots)
- Bibliometrics (citations, author metrics)
August 27, 2025 at 6:31 PM
We trained models on ~15,000 journals labeled by DOAJ (12,869 legitimate vs 2,536 removed). Features included:
- Website content (editorial boards, policies)
- Website design (HTML structure, screenshots)
- Bibliometrics (citations, author metrics)
- Website content (editorial boards, policies)
- Website design (HTML structure, screenshots)
- Bibliometrics (citations, author metrics)
🚨New paper🚨
Open access has expanded science’s reach but also fueled the rise of "questionable" journals. Manual vetting can’t keep pace with thousands of titles and bad actors who adapt quickly. Wrong incentives too strong.
In a new Science Advances paper we ask: can AI help?
Open access has expanded science’s reach but also fueled the rise of "questionable" journals. Manual vetting can’t keep pace with thousands of titles and bad actors who adapt quickly. Wrong incentives too strong.
In a new Science Advances paper we ask: can AI help?
August 27, 2025 at 6:31 PM
🚨New paper🚨
Open access has expanded science’s reach but also fueled the rise of "questionable" journals. Manual vetting can’t keep pace with thousands of titles and bad actors who adapt quickly. Wrong incentives too strong.
In a new Science Advances paper we ask: can AI help?
Open access has expanded science’s reach but also fueled the rise of "questionable" journals. Manual vetting can’t keep pace with thousands of titles and bad actors who adapt quickly. Wrong incentives too strong.
In a new Science Advances paper we ask: can AI help?