Explaining Latent Factor Models for Recommendation with Influence Functions

TL;DR

This paper introduces FIA, a fast influence analysis method that provides explicit neighbor-style explanations for latent factor models in recommendation systems, improving interpretability without sacrificing efficiency.

Contribution

The paper develops a novel influence function-based approach for explaining LFMs, including an efficient algorithm for matrix factorization and an extension to neural collaborative filtering.

Findings

FIA delivers accurate neighbor-style explanations for LFMs.

The influence computation algorithm is highly efficient and scalable.

Experimental results confirm the method's correctness and usefulness.

Abstract

Latent factor models (LFMs) such as matrix factorization achieve the state-of-the-art performance among various Collaborative Filtering (CF) approaches for recommendation. Despite the high recommendation accuracy of LFMs, a critical issue to be resolved is the lack of explainability. Extensive efforts have been made in the literature to incorporate explainability into LFMs. However, they either rely on auxiliary information which may not be available in practice, or fail to provide easy-to-understand explanations. In this paper, we propose a fast influence analysis method named FIA, which successfully enforces explicit neighbor-style explanations to LFMs with the technique of influence functions stemmed from robust statistics. We first describe how to employ influence functions to LFMs to deliver neighbor-style explanations. Then we develop a novel influence computation algorithm for matrix factorization with high efficiency. We further extend it to the more general neural collaborative filtering and introduce an approximation algorithm to accelerate influence analysis over neural network models. Experimental results on real datasets demonstrate the correctness, efficiency and usefulness of our proposed method.