Hyeon-woo Kim, Yong-jun Kim, Gil-sang Yoo, Journal of Internet Computing and Services, Vol. 25, No. 1, pp. 69-78, Feb. 2024
10.7472/jksii.2024.25.1.69, Full Text:  HTML
Keywords: MultiSAGE, recommend system, ESG, GraphSAGE

Abstract

Recently, consumers have shown an increasing tendency to seek information related to environmental, social, and governance (ESG) aspects in order to choose products with higher social value and environmental friendliness. In this paper, we proposes a product recommendation system applying ESG indicators tailored to the recent consumer trend of value-based consumption, utilizing a model called MultiSAGE that combines GraphSAGE and GAT. To achieve this, ESG rating data for 1,033 companies in 2022 collected from the Korea ESG Standard Institute and actual product data from N companies were transformed into a Heterogeneous Graph format through a data processing pipeline. The MultiSAGE model was then applied in machine learning to implement a recommendation system that, given a specific product, suggests eco-friendly alternatives. The implementation results indicate that consumers can easily compare and purchase products with ESG indicators applied, and it is anticipated that this system will be utilized in recommending products with social value and environmental friendliness.

Statistics
Cite this article