Interlock-Free Multi-Aspect Rationalization for Text Classification

TL;DR

This paper introduces a multi-stage training approach with contrastive loss to generate diverse, meaningful rationales for multi-aspect text classification, addressing interlocking issues in rationalization.

Contribution

It proposes a novel multi-stage training method with contrastive loss to improve multi-aspect rationalization, overcoming interlocking problems in existing frameworks.

Findings

Significant improvement in rationalization performance on beer review dataset

Enhanced semantic diversity of generated rationales

Effective handling of interlocking dynamics in multi-aspect rationalization

Abstract

Explanation is important for text classification tasks. One prevalent type of explanation is rationales, which are text snippets of input text that suffice to yield the prediction and are meaningful to humans. A lot of research on rationalization has been based on the selective rationalization framework, which has recently been shown to be problematic due to the interlocking dynamics. In this paper, we show that we address the interlocking problem in the multi-aspect setting, where we aim to generate multiple rationales for multiple outputs. More specifically, we propose a multi-stage training method incorporating an additional self-supervised contrastive loss that helps to generate more semantically diverse rationales. Empirical results on the beer review dataset show that our method improves significantly the rationalization performance.