[Retracted] Cross-Domain End-To-End Aspect-Based Sentiment Analysis with Domain-Dependent Embeddings

Figure 6

The change of F1 scores using the word embedding layer and glove embeddings in CD-E2EABSA after 4 epochs. (i) CD-E2EABSA (S) uses Bert as the embedding layer for thw source domain; (ii) CD-E2EABSA (T) uses Bert as the embedding layer for the target domain; (iii) CD-E2EABSA (Glove) (S) uses glove embedding as input for the source domain; and (iv) CD-E2EABSA (Glove) (T) uses glove embedding as input for the target domain.