Research Area:  Machine Learning
Aspect-based sentiment analysis (ABSA), which aims to identify fine-grained opinion polarity towards a specific aspect, is a challenging subtask of sentiment analysis (SA). In this paper, we construct an auxiliary sentence from the aspect and convert ABSA to a sentence-pair classification task, such as question answering (QA) and natural language inference (NLI). We fine-tune the pre-trained model from BERT and achieve new state-of-the-art results on SentiHood and SemEval-2014 Task 4 datasets.
Author(s) Name:  Chi Sun, Luyao Huang, Xipeng Qiu
Journal name:  Computer Science
Publisher name:  arXiv:1903.09588
Paper Link:   https://arxiv.org/abs/1903.09588