The Limits of Graph Samplers for Training Inductive Recommender Systems: Extended results

TL;DR

This paper investigates the effectiveness of graph sampling techniques in training inductive recommender systems, revealing trade-offs between data reduction, training time, and recommendation performance, and emphasizing the importance of temporal data considerations.

Contribution

It provides an empirical evaluation of graph sampling methods for inductive recommender systems and highlights the need for new sampling techniques that incorporate temporal information.

Findings

Sampling with 50% of data maintains performance and reduces training time by up to 86%.

Reducing training data further degrades recommendation quality.

Temporal aspects are crucial for effective graph sampling in recommendation tasks.

Abstract

Inductive Recommender Systems are capable of recommending for new users and with new items thus avoiding the need to retrain after new data reaches the system. However, these methods are still trained on all the data available, requiring multiple days to train a single model, without counting hyperparameter tuning. In this work we focus on graph-based recommender systems, i.e., systems that model the data as a heterogeneous network. In other applications, graph sampling allows to study a subgraph and generalize the findings to the original graph. Thus, we investigate the applicability of sampling techniques for this task. We test on three real world datasets, with three state-of-the-art inductive methods, and using six different sampling methods. We find that its possible to maintain performance using only 50% of the training data with up to 86% percent decrease in training time; however, using less training data leads to far worse performance. Further, we find that when it comes to data for recommendations, graph sampling should also account for the temporal dimension. Therefore, we find that if higher data reduction is needed, new graph based sampling techniques should be studied and new inductive methods should be designed.