Learning Intrinsic Dimension via Information Bottleneck for Explainable Aspect-based Sentiment Analysis
This work addresses the need for more accurate and interpretable explanations in ABSA, which is important for users in NLP applications, though it is incremental as it builds on existing gradient-based methods.
The paper tackles the problem of improving gradient-based explanation methods for Aspect-based Sentiment Analysis (ABSA) by addressing the assumption of equal significance across all gradient dimensions, proposing an Information Bottleneck-based Gradient (IBG) framework that refines embeddings into an intrinsic dimension to enhance model performance and interpretability.
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 (\texttt{IBG}) 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 \texttt{IBG} approach considerably improves both the models' performance and interpretability by identifying sentiment-aware features.