Explaining the Deep Natural Language Processing by Mining Textual Interpretable Features

TL;DR

This paper introduces T-EBAnO, a new explainability framework for deep NLP models that extracts interpretable features and quantifies their influence to make model decisions more transparent and human-readable.

Contribution

The paper presents T-EBAnO, a novel framework that provides local and global explanations for deep NLP models using new influence indexes and interpretable feature extraction.

Findings

Effective explanations for BERT in sentiment analysis

Accurate interpretation of LSTM in toxic comment classification

Demonstrated improved transparency of deep NLP models

Abstract

Despite the high accuracy offered by state-of-the-art deep natural-language models (e.g. LSTM, BERT), their application in real-life settings is still widely limited, as they behave like a black-box to the end-user. Hence, explainability is rapidly becoming a fundamental requirement of future-generation data-driven systems based on deep-learning approaches. Several attempts to fulfill the existing gap between accuracy and interpretability have been done. However, robust and specialized xAI (Explainable Artificial Intelligence) solutions tailored to deep natural-language models are still missing. We propose a new framework, named T-EBAnO, which provides innovative prediction-local and class-based model-global explanation strategies tailored to black-box deep natural-language models. Given a deep NLP model and the textual input data, T-EBAnO provides an objective, human-readable, domain-specific assessment of the reasons behind the automatic decision-making process. Specifically, the framework extracts sets of interpretable features mining the inner knowledge of the model. Then, it quantifies the influence of each feature during the prediction process by exploiting the novel normalized Perturbation Influence Relation index at the local level and the novel Global Absolute Influence and Global Relative Influence indexes at the global level. The effectiveness and the quality of the local and global explanations obtained with T-EBAnO are proved on (i) a sentiment analysis task performed by a fine-tuned BERT model, and (ii) a toxic comment classification task performed by an LSTM model.