Improving Interpretability via Explicit Word Interaction Graph Layer

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.

Authors

• Arshdeep Sekhon University of Virginia
• Hanjie Chen University of Virginia
• Aman Shrivastava University of Virginia
• Zhe Wang Griffith University
• Yangfeng Ji University of Virginia
• Yanjun Qi University of Virginia

Keywords

• ML: Graph-based Machine Learning
• SNLP: Interpretability & Analysis of NLP Models