Revisiting Graph Projections for Effective Complementary Product Recommendation

TL;DR

This paper revisits bipartite graph projections to improve complementary product recommendations, demonstrating significant performance gains over existing methods despite its simplicity.

Contribution

It introduces a novel approach for inferring product complementarity from user-item bipartite graphs, enhancing recommendation accuracy.

Findings

Average improvement of +43% over sequential recommenders

Average improvement of +38% over graph-based recommenders

Effective handling of noisy and sparse user-item interactions

Abstract

Complementary product recommendation is a powerful strategy to improve customer experience and retail sales. However, recommending the right product is not a simple task because of the noisy and sparse nature of user-item interactions. In this work, we propose a simple yet effective method to predict a list of complementary products given a query item, based on the structure of a directed weighted graph projected from the user-item bipartite graph. We revisit bipartite graph projections for recommender systems and propose a novel approach for inferring complementarity relationships from historical user-item interactions. We compare our model with recent methods from the literature and show, despite the simplicity of our approach, an average improvement of +43% and +38% over sequential and graph-based recommenders, respectively, over different benchmarks.