Interpreting Deep Learning Models in Natural Language Processing: A Review

TL;DR

This paper reviews various interpretation methods for neural network models in NLP, highlighting their categories, sub-categories, limitations, and future research directions to improve model interpretability.

Contribution

It provides a comprehensive taxonomy and detailed analysis of existing interpretation methods for neural NLP models, identifying gaps and proposing future research avenues.

Findings

High-level taxonomy of interpretation methods in NLP

Detailed description of sub-categories like influence functions and attention

Identification of deficiencies and future research directions

Abstract

Neural network models have achieved state-of-the-art performances in a wide range of natural language processing (NLP) tasks. However, a long-standing criticism against neural network models is the lack of interpretability, which not only reduces the reliability of neural NLP systems but also limits the scope of their applications in areas where interpretability is essential (e.g., health care applications). In response, the increasing interest in interpreting neural NLP models has spurred a diverse array of interpretation methods over recent years. In this survey, we provide a comprehensive review of various interpretation methods for neural models in NLP. We first stretch out a high-level taxonomy for interpretation methods in NLP, i.e., training-based approaches, test-based approaches, and hybrid approaches. Next, we describe sub-categories in each category in detail, e.g., influence-function based methods, KNN-based methods, attention-based models, saliency-based methods, perturbation-based methods, etc. We point out deficiencies of current methods and suggest some avenues for future research.