Related Experiment Video
Updated: Jun 11, 2026
07:35
Selecting Multiple Biomarker Subsets with Similarly Effective Binary Classification Performances
Published on: October 11, 2018
The Deep Learning-Based Recommender System "Pubmender" for Choosing a Biomedical Publication Venue: Development and
Xiaoyue Feng1,2, Hao Zhang1, Yijie Ren2
1Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education, College of Computer Science and Technology, Jilin University, Changchun, China.
Journal of Medical Internet Research
|May 26, 2019
Summary
A new publication recommender system, Pubmender, uses deep learning to suggest suitable PubMed journals based on paper abstracts. This tool significantly improves journal recommendation accuracy for biomedical research.
Area of Science:
- Biomedical Informatics
- Computational Biology
- Scientific Publishing
Background:
- Choosing high-quality publication venues is crucial but challenging for researchers due to the expanding number of journals.
- Existing recommender systems have not adequately addressed the need for publication venue recommendations, particularly within biomedical research.
- No dedicated recommender system exists for suggesting journals indexed in PubMed, the largest biomedical literature database.
Purpose of the Study:
- To develop and evaluate Pubmender, a novel publication recommender system designed to suggest appropriate PubMed journals based on a research paper's abstract.
- To provide a specialized tool for biomedical scientists and clinicians to identify suitable publication venues.
Main Methods:
- Utilized pretrained word2vec for initial feature space construction.
- Developed a deep convolutional neural network (CNN) to generate high-level abstract representations.
- Employed a fully connected softmax model for journal recommendation.
Main Results:
- The system achieved superior accuracy in top 10 recommendations compared to existing tools like Microsoft Academic Search (MAS), ACM, and CiteSeer.
- Pubmender demonstrated significantly higher accuracy (329% and 406% greater) than Elsevier's Journal Finder and Springer's Journal Suggester, respectively.
- The performance was validated using a dataset of 880,165 papers from 1130 PubMed Central journals.
Conclusions:
- The developed deep learning-based recommender system effectively suggests relevant journals for publication.
- Pubmender offers a valuable resource for biomedical researchers and clinicians to select optimal venues for their manuscripts.
- The system is freely accessible as a web service.
