JASTE: An Efficient Multi-Task Model for Aspect Sentiment Triplet Extraction

Aspect sentiment triplet extraction (ASTE) is a relative difficult and novel research, which subtask of aspect-based analysis (ABSA). ASTE task that extracts triplets aspects being discussed, relevant opinion entities polarities from given sentence. Existing approaches mainly deal with this problem by pipeline or simple multi-task structure, do not take full advantage the strong correlation among three elements triplet. In work, we adopt two special tagging schemes, AOBIO Pair Tagging Scheme (PTS), propose an efficient end-to-end model named Joint Sentiment Triplet Extraction (JASTE) to address task. JASTE composed modules: aspect module, relation module module. Specially, designed capture relationship between properly. The modules interact each other sharing same embedding. Extensive experimental results on different benchmark datasets show can significantly outperform state-of-the-art performances.