CLEAR: Causal Explanations from Attention in Neural Recommenders

TL;DR

CLEAR is a novel method that derives session-specific causal graphs from attention mechanisms in neural recommenders, enabling counterfactual explanations for user behavior and recommendations.

Contribution

The paper introduces CLEAR, a new approach for extracting causal explanations from attention in recommenders, even with latent confounders, enhancing interpretability.

Findings

Counterfactual explanations are shorter and more informative.

Alternative recommendations are ranked higher with CLEAR.

CLEAR outperforms naive attention-based explanations.

Abstract

We present CLEAR, a method for learning session-specific causal graphs, in the possible presence of latent confounders, from attention in pre-trained attention-based recommenders. These causal graphs describe user behavior, within the context captured by attention, and can provide a counterfactual explanation for a recommendation. In essence, these causal graphs allow answering "why" questions uniquely for any specific session. Using empirical evaluations we show that, compared to naively using attention weights to explain input-output relations, counterfactual explanations found by CLEAR are shorter and an alternative recommendation is ranked higher in the original top-k recommendations.