Early indicators of scientific impact: Predicting citations with altmetrics
作者：
Highlights：
• We described a machine learning approach to predict citations using altmetrics.
• We built and tested several classifiers and regressors.
• We found that the neural networks and ensemble models performed better than other models.
• We present the predictions for short-term and long-term scholarly impact.
• Our experiments showed some new and important factors in predicting citations.
摘要
•We described a machine learning approach to predict citations using altmetrics.•We built and tested several classifiers and regressors.•We found that the neural networks and ensemble models performed better than other models.•We present the predictions for short-term and long-term scholarly impact.•Our experiments showed some new and important factors in predicting citations.
论文关键词：Citation count,Citation prediction,Altmetrics,Scientometrics,Scholarly communication,Social media,Science of science,Scholarly impact,Metascience
论文评审过程：Received 30 December 2018, Revised 17 December 2020, Accepted 21 December 2020, Available online 11 February 2021, Version of Record 11 February 2021.
论文官网地址：https://doi.org/10.1016/j.joi.2020.101128
