Online recommendation based on incremental-input self-organizing map
作者：
Highlights：
• We change the basic structure of the input layer with a fixed number of units, which is usually adopted by most neural networks, and develop an incremental-input SOM whose input layer can be “expanded” by adding new input units.
• We propose an online recommendation algorithm based on the incremental-input SOM and provide two different updating mechanisms, which can handle the arrivals of new consumers and new products simultaneously.
• The algorithm is superior to several existing online recommendation algorithms, and performs similarly to the static recommendation algorithm which need to be retrained by all of the data.
摘要
•We change the basic structure of the input layer with a fixed number of units, which is usually adopted by most neural networks, and develop an incremental-input SOM whose input layer can be “expanded” by adding new input units.•We propose an online recommendation algorithm based on the incremental-input SOM and provide two different updating mechanisms, which can handle the arrivals of new consumers and new products simultaneously.•The algorithm is superior to several existing online recommendation algorithms, and performs similarly to the static recommendation algorithm which need to be retrained by all of the data.
论文关键词：Self-organizing map,Incremental input,Incremental learning,Online recommendation
论文评审过程：Received 22 February 2021, Revised 6 September 2021, Accepted 22 September 2021, Available online 24 September 2021, Version of Record 15 November 2021.
论文官网地址：https://doi.org/10.1016/j.elerap.2021.101096
