Dual-Channel Span for Aspect Sentiment Triplet Extraction

Aspect Sentiment Triplet Extraction (ASTE) is one of the compound tasks of fine-grained aspect-based sentiment analysis (ABSA), aiming at extracting the triplets of aspect terms, corresponding opinion terms and the associated sentiment orientation. Recent efforts in exploiting span-level semantic interaction have shown superior performance on ASTE task. However, span-based approaches could suffer from excessive noise due to the large number of spans that have to be considered. To ease this burden, we propose a dual-channel span generation method to coherently constrain the search space of span candidates. Specifically, we leverage the syntactic relations among aspect/opinion terms and their part-of-speech characteristics to generate useful span candidates, which empirically reduces span enumeration by nearly a half. Besides, the interaction between syntactic and part-of-speech views brings relevant linguistic information to learned span representations. Extensive experiments on two public datasets demonstrate both the effectiveness of our design and the superiority on ASTE task.