Graph-based Recommendation for Sparse and Heterogeneous User Interactions

TL;DR

A graph-based recommender model is proposed which utilizes heterogeneous interactions between users and content of different types and is able to operate well on small-scale datasets.

Abstract

Recommender system research has oftentimes focused on approaches that operate on large-scale datasets containing millions of user interactions. However, many small businesses struggle to apply state-of-the-art models due to their very limited availability of data. We propose a graph-based recommender model which utilizes heterogeneous interactions between users and content of different types and is able to operate well on small-scale datasets. A genetic algorithm is used to find optimal weights that represent the strength of the relationship between users and content. Experiments on two real-world datasets (which we make available to the research community) show promising results (up to 7% improvement), in comparison with other state-of-the-art methods for low-data environments. These improvements are statistically significant and consistent across different data samples.

Figures (3)

• Figure 1: Example of the heterogeneous graph in a social network.
• Figure 2: MRR@k for varying choices of the cutoff threshold $k$.
• Figure 3: Plot of how the outgoing edge weights evolve for different sizes of the insurance dataset.