Disease profiling in pharmaceutical E-commerce

作者：

Highlights：

• Unlabeled purchase data are turned into partially labeled data with heuristics.

• A prudent iterative naïve Bayesian (PINB) algorithm is proposed.

• PINB demonstrates high performance in predicting users’ disease label.

摘要

•Unlabeled purchase data are turned into partially labeled data with heuristics.•A prudent iterative naïve Bayesian (PINB) algorithm is proposed.•PINB demonstrates high performance in predicting users’ disease label.

论文关键词：User profile,Bayesian classifier,Semi-supervised learning,Pharmaceutical e-commerce

论文评审过程：Received 1 July 2020, Revised 15 March 2021, Accepted 7 April 2021, Available online 13 April 2021, Version of Record 21 April 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115015