Improving Interpretability via Explicit Word Interaction Graph Layer

TL;DR

This paper introduces WIGRAPH, a trainable neural network layer that learns word interaction graphs to improve interpretability and prediction accuracy of NLP models, applicable across various architectures and datasets.

Contribution

The paper presents WIGRAPH, a novel layer that explicitly models word interactions to enhance interpretability and performance in neural NLP classifiers.

Findings

WIGRAPH improves interpretability of NLP models.

Adding WIGRAPH enhances prediction accuracy.

WIGRAPH is compatible with multiple NLP architectures.

Abstract

Recent NLP literature has seen growing interest in improving model interpretability. Along this direction, we propose a trainable neural network layer that learns a global interaction graph between words and then selects more informative words using the learned word interactions. Our layer, we call WIGRAPH, can plug into any neural network-based NLP text classifiers right after its word embedding layer. Across multiple SOTA NLP models and various NLP datasets, we demonstrate that adding the WIGRAPH layer substantially improves NLP models' interpretability and enhances models' prediction performance at the same time.