.png){kind=icon}
.png){kind=icon}
Return
Online recommendation based on incremental-input self-organizing map
.png){kind=icon}
PRE
AI
DOI：10.1016/j.elerap.2021.101096.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
E-commerce platforms usually use recommendation algorithms to help consumers find products that they may prefer. Most traditional recommendation models, which are generally built on the basis of static data, may fail in managing newly arrived ratings of new consumers or on new products. These static recommendation models need to be retrained offline using the newly arrived ratings, which is often inefficient and time consuming. In this paper, we propose a novel incremental-input self-organizing map (SOM) and apply it to generate an online recommendation model. We add new consumers into the recommendation model by augmenting the input units and updating the corresponding weights between the input layer and the output layer. We treat new products as new samples arriving at the SOM and specially design an incremental learning strategy for SOM. Thus, the proposed model does not have to be retrained and can be updated online with the very few ratings of new consumers or new products. Experimental results on real-world datasets indicate that the proposed algorithm is superior to several existing online recommendation algorithms in terms of recommendation accuracy and performs similarly to the static recommendation algorithm which needs to be retrained by all of the data.
Keywords:
Self-organizing map
Incremental input
Incremental learning
Online recommendation
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
6.3
Papers:
2.4K
Citations:
5.9K