D-Separation for Causal Self-Explanation

TL;DR

This paper introduces a novel causal rationalization method for NLP models that uses the Minimum Conditional Dependence criterion based on D-separation, improving interpretability and accuracy over traditional mutual information approaches.

Contribution

The paper proposes the MCD criterion grounded in D-separation to better identify causal rationales, addressing limitations of MMI-based methods.

Findings

MCD improves F1 score by up to 13.7% over state-of-the-art methods.

Empirical validation shows MCD effectively uncovers causal rationales.

The approach enhances interpretability and robustness of NLP models.

Abstract

Rationalization is a self-explaining framework for NLP models. Conventional work typically uses the maximum mutual information (MMI) criterion to find the rationale that is most indicative of the target label. However, this criterion can be influenced by spurious features that correlate with the causal rationale or the target label. Instead of attempting to rectify the issues of the MMI criterion, we propose a novel criterion to uncover the causal rationale, termed the Minimum Conditional Dependence (MCD) criterion, which is grounded on our finding that the non-causal features and the target label are \emph{d-separated} by the causal rationale. By minimizing the dependence between the unselected parts of the input and the target label conditioned on the selected rationale candidate, all the causes of the label are compelled to be selected. In this study, we employ a simple and practical measure of dependence, specifically the KL-divergence, to validate our proposed MCD criterion. Empirically, we demonstrate that MCD improves the F1 score by up to $13.7\%$ compared to previous state-of-the-art MMI-based methods. Our code is available at: \url{https://github.com/jugechengzi/Rationalization-MCD}.