Identifying breakthrough scientific papers

作者：

Highlights：

• An LDA topic model and SVM classifier for predicting paper publication years have been trained.

• A real-number paper innovation score based on publication year prediction error is proposed.

• High citation counts usually correlate to high innovation scores.

摘要

•An LDA topic model and SVM classifier for predicting paper publication years have been trained.•A real-number paper innovation score based on publication year prediction error is proposed.•High citation counts usually correlate to high innovation scores.

论文关键词：Scientometrics,Document timestamping,Topic models,Classification

论文评审过程：Received 21 April 2019, Revised 8 October 2019, Accepted 11 November 2019, Available online 22 November 2019, Version of Record 22 November 2019.

论文官网地址：https://doi.org/10.1016/j.ipm.2019.102168