An Analysis of Attention Mechanisms: The Case of Word Sense Disambiguation in Neural Machine Translation
This provides insights into how NMT models handle word sense disambiguation, which is incremental as it clarifies existing mechanisms rather than proposing new ones.
The paper analyzed encoder-decoder attention mechanisms in neural machine translation for word sense disambiguation, finding that attention distributes more to ambiguous nouns themselves rather than context tokens, suggesting it is not the main mechanism for incorporating contextual information.
Recent work has shown that the encoder-decoder attention mechanisms in neural machine translation (NMT) are different from the word alignment in statistical machine translation. In this paper, we focus on analyzing encoder-decoder attention mechanisms, in the case of word sense disambiguation (WSD) in NMT models. We hypothesize that attention mechanisms pay more attention to context tokens when translating ambiguous words. We explore the attention distribution patterns when translating ambiguous nouns. Counter-intuitively, we find that attention mechanisms are likely to distribute more attention to the ambiguous noun itself rather than context tokens, in comparison to other nouns. We conclude that attention mechanism is not the main mechanism used by NMT models to incorporate contextual information for WSD. The experimental results suggest that NMT models learn to encode contextual information necessary for WSD in the encoder hidden states. For the attention mechanism in Transformer models, we reveal that the first few layers gradually learn to "align" source and target tokens and the last few layers learn to extract features from the related but unaligned context tokens.