AiPromptss
Back to browse

Generate a title for a paper to be published in a top journal

I will provide you with the abstract and key words of a scientific paper in any language and you will detect the language and reply in the same language. Your…

Added May 19, 20260 views0 copies
Prompt
I will provide you with the abstract and key words of a scientific paper in any language and you will detect the language and reply in the same language. Your task is to provide me with the title of the scientific paper based on the abstract and key words in thevsame language. The title of the scientific paper should be concise, clear and informative. You should avoid using wasted words such as "a study of," "investigation of," "development of," or"observations on." Make sure the title can grip the audience immediately. My abstract is"事件抽取旨在将非结构化自然语言文本中的事件信息以结构化形式进行识别提取。传统事件抽取方法抽取范围局限
于单个句子,且依赖较大规模的标注数据,在篇章级抽取任务与低资源目标领域上性能不足。现有研究利用提示学习方法,以模板槽位填空方式实现篇章级事件抽取,其缺点在于传统提示模板槽位对论元角色分类准确度不足,容易造成论元角色抽取错误。针对上述问题,提出一种基于槽位语义增强提示学习的篇章级事件抽取方法,在提示学习方法的基础上,将传统事件抽取范式中的论元角色语义信息融入提示模板槽位中,为模型的槽位预测生成环节提供论元类型约束,提高篇章级事件抽取准确率。通过使预训练语言模型上下游任务保持一致,模型具有更好的泛化能力,并能够以较低成本实现知识迁移,在低资源事件抽取场景下取得了更好的性能。实验结果表明,相较于传统基线方法,该方法在包含 59 种论元类型的英文事件抽取数据集上获得了 2.6 的 F1 提升,在包含 92 类论元类型的中文数据集上获得了 2.9 的 F1 提升,并在低资源数据规模下获得了4.0 的 F1 提升。" My key words are "事件抽取;提示学习;信息抽取;自然语言处理;预训练语言模型"

Replace text in [BRACKETS] with your own values before pasting.