Heat Up The Sentiment Learning With Ice

Yao Yao, Zuchao Li, hai zhao

Abstract: Recently, dramatic gains have been made on the task of aspect sentiment triplet extraction (ASTE). In this paper, we introduce a straightforward pipeline model to perform two-stage sequence labeling, including aspect and opinion terms identification and aspect-opinion pair classification. To exploit the cross-sentence context information to the maximum extent possible, we propose the instance cooperative enhancement (ICE) by introducing unsupervised clustering methods. Through experimenting with various clustering methods, we found that GSDMM unleashes the potential of cross-sentence information to the most degree. Compared to current state-of-the-art models, the results show the effectiveness of our proposed framework on ASTE-Data-V2.