Aspect-Based Pair-Wise Opinion Generation in Chinese automotive reviews: Design of the task, dataset and model
作者：
Highlights：
• We introduce a new task of pair–wise opinion generation based on reviews and aspects.
• We construct a large–scale multiaspect dataset, which can be used as a benchmark.
• We propose a novel model to generate opinion target–word jointly for a given aspect.
• The proposed method outperforms several strong baselines and keeps interpretability.
• We conduct detailed analyses for better understanding the data and model.
摘要
•We introduce a new task of pair–wise opinion generation based on reviews and aspects.•We construct a large–scale multiaspect dataset, which can be used as a benchmark.•We propose a novel model to generate opinion target–word jointly for a given aspect.•The proposed method outperforms several strong baselines and keeps interpretability.•We conduct detailed analyses for better understanding the data and model.
论文关键词：Fine-grained opinion analysis,Aspect-based generation,Pair-wise generation,Joint model
论文评审过程：Received 1 March 2021, Revised 12 August 2021, Accepted 12 August 2021, Available online 14 September 2021, Version of Record 14 September 2021.
论文官网地址：https://doi.org/10.1016/j.ipm.2021.102729
