ASOTE-Data

ASOTE-Data是由深圳大学的李云聪等人创建的数据集，专注于方面情感三元组提取（ASTE）。该数据集不仅包含句子中的三元组（方面术语、情感、观点术语），还额外包括了没有三元组的句子及不属于任何三元组的方面术语，使得数据集更符合真实世界场景。ASOTE-Data通过三个步骤进行标注：首先标注句子中的方面术语，其次标注这些方面术语对应的观点术语，最后标注方面术语与观点术语对的情感。此数据集适用于评估模型在真实场景中的表现，并已证明能有效提升模型的泛化性能。

ASOTE-Data is a dataset developed by Li Yuncong et al. from Shenzhen University, focusing on Aspect Sentiment Triplet Extraction (ASTE). This dataset not only contains the triplets (aspect term, sentiment, opinion term) in sentences, but also additionally includes sentences without any triplets and aspect terms that do not belong to any triplet, making the dataset more aligned with real-world scenarios. ASOTE-Data adopts a three-step annotation workflow: first, annotate the aspect terms in the target sentences; second, annotate the opinion terms corresponding to these aspect terms; finally, annotate the sentiment polarity of each aspect-opinion term pair. This dataset is applicable for evaluating the real-world performance of models, and it has been proven to effectively improve the generalization performance of models.