Does BERT Look at Sentiment Lexicon?

Abstract

AbstractThe main approaches to sentiment analysis are rule-based methods and machine learning, in particular, deep neural network models with the Transformer architecture, including BERT. The performance of neural network models in the tasks of sentiment analysis is superior to the performance of rule-based methods. The reasons for this situation remain unclear due to the poor interpretability of deep neural network models. One of the main keys to understanding the fundamental differences between the two approaches is the analysis of how sentiment lexicon is taken into account in neural network models. To this end, we study the attention weights matrices of the Russian-language RuBERT model. We fine-tune RuBERT on sentiment text corpora and compare the distributions of attention weights for sentiment and neutral lexicons. It turns out that, on average, 3/4 of the heads of various model variants statistically pay more attention to the sentiment lexicon compared to the neutral one.KeywordsSentiment analysisSentiment lexiconsBERTInterpretable modelsAttention