First, the shift towards ML, especially supervised and/or DL solutions for ABSA, highlights the importance of dataset resources. In particular, annotated large benchmark datasets are crucial for the quality and development of ABSA research. Meanwhile, the finer granularity of ABSA also brings the p...
Secondly, RoBERTa removes the next sentence prediction task from BERT, which has little impact on performance. Finally, RoBERTa expands the batch size and word list, allowing the model to use a larger dataset during pre-training, resulting in richer semantic information at the end of pre-...