中科院之声|试试让机器自己生成摘要?丨智言智语( 二 )


相关实验表明 , 融合翻译模式的跨语言自动摘要方法能够生成与基于多任务学习方法质量相当的摘要 , 但相比之下前者具有降低模型对于数据的依赖、减小模型容量和提升训练效率的优势 。
参考文献:
1. Junnan Zhu, Yu Zhou, JiajunZhang, and Chengqing Zong. Attend, Translate and Summarize: An Efficient Methodfor Neural Cross-Lingual Summarization. In Proceedings of the 58th AnnualMeeting of the Association for Computational Linguistics (ACL), Online, July5-July 10, 2020, pp. 1309-1321.
2. Junnan Zhu, Qian Wang, YiningWang, Yu Zhou, Jiajun Zhang, Shaonan Wang, and Chengqing Zong. NCLS: NeuralCross-Lingual Summarization. In Proceedings of 2019 Conference on EmpiricalMethods in Natural Language Processing and 9th International Joint Conferenceon Natural Language Processing (EMNLP- IJCNLP), November 3–7, Hong Kong, China,2019, pp.3045-3055.
配图:董晓芙


推荐阅读