Aspect-based sentiment analysis method using text generation
Ke Yan, Lianggui Tang, Meilin Wu, Qingda Zhang, Xiuling Zhu · 2023
Aspect-Based Sentiment Analysis (ABSA) aims to identify aspect terms, corresponding sentiment polarities, and opinion terms. Most studies only focus on a subset of these tasks, which leads to various complicated ABSA models. We propose a text-based generation framework for various ABSA tasks. It uses a sequence-to-sequence text generation method that can selectively output aspect terms, sentiment polarity, and corresponding opinion items of input sentences, than normalize the results using glove word vectors. Extensive experiments show the effectiveness of our proposed approach,achieves state-of-the-art performance on multiple datasets.