Aspect-level sentiment analysis model incorporating part-of-speech
Lei Wu, Tengbin Wang · 2022
Aspect-level sentiment analysis aims to identify the sentiment polarity of various aspects in the text. Existing deep learning models generally use the syntactic dependency tree as the model input when extracting the grammatical information of the text. In this paper, we argue that the part-of-speech sequence of the text carries certain grammatical information. We concatenate the word vector and the part-of-speech vector and inputs it into Bi-directional Long Short-Term Memory (Bi- LSTM), and then combined the aspect-specific masking method and attention mechanism for sentiment analysis. Experimental results on SemEval 2014 Datasets demonstrate the effectiveness of our model.