Research on aspect-level sentiment analysis of tourism review text based on joint learning
Shao Yu-fei, Du Lei · 2023
Aiming at the problem that the current aspect level sentiment analysis model is weak in generalization ability and word perception, a multi-task joint learning model GL-IJM for target word emotional polarity discrimination and target word extraction is proposed, which can learn the relationship between specific aspects and context in the training process, and can be used to capture the information closely related to specific aspects in the context sequence, so as to increase the modeling of specific aspects, compared with the BERT-based method. The F1 values of ATE tasks on camera and mobile phone datasets increased by 0.2 and 0.4 percentage points, respectively, and the F1 values of SPC tasks increased by 0.61 and 0.04 percentage points, respectively.