In-Processing and Post-Processing Strategies for Balancing Accuracy and Sustainability in Product Recommendations
(2023)
Files
No attached file found for this publication.
Details
- Authors
Satinet, Chloé 
- Abstract
- (en) With the increasing awareness around the environmental impact of our consumption, recommender systems, well-known for encouraging purchases and consumption, are contested. Consequently, in recent literature, it has been suggested for recommender systems to balance exploitation of existing preferences and exploration of sustainable items, i.e., to make sustainable alternatives to conventional products more accessible to consumers and enable sustainable consumption habits. In this paper, we therefore analyse how to increase the presence of less-popular sustainable products in recommendation lists, without overly decreasing the accuracy of the recommendations. More precisely, we test three in-processing and four post-processing strategies using an offline experimental design. Some strategies manage to offer interesting accuracy-sustainability trade-offs. For instance, when applying a post-processing strategy relying on a calibration framework on a content-based recommendation algorithm, we have up to 20% sustainability gain (represented by the decrease in the average environmental impact of the recommended products) without accuracy loss. Higher sustainability improvements can be achieved if a loss of accuracy is tolerated.
- Affiliations
Citations
Satinet, C., Fouss, F., Saerens, M., & Leleux, P. (2023). In-Processing and Post-Processing Strategies for Balancing Accuracy and Sustainability in Product Recommendations (Louvain Research Institute in Management and Organizations Working Paper Series). https://hdl.handle.net/2078.5/217240