Learning Intrinsic Dimension via Information Bottleneck for Explainable Aspect-based Sentiment Analysis

Gradient-based explanation methods are increasingly used to interpret neural models in natural language processing (NLP) due to their high fidelity. Such methods determine word-level importance using dimension-level gradient values through a norm function, often presuming equal significance for all gradient dimensions. However, in the context of Aspect-based Sentiment Analysis (ABSA), our preliminary research suggests that only specific dimensions are pertinent. To address this, we propose the Information Bottleneck-based Gradient (textttIBG) explanation framework for ABSA. This framework leverages an information bottleneck to refine word embeddings into a concise intrinsic dimension, maintaining essential features and omitting unrelated information. Comprehensive tests show that our textttIBG approach considerably improves both the models' performance and interpretability by identifying sentiment-aware features.