【摘要】 针对现有的预训练模型难以进行微调以及生成式摘要算法容易产生未登录词的问题提出一种基于关键词抽取和提示学习的摘要方法。首先添加提示学习(Prompt Learning)通过向输入增加提示信息，将下游任务改成文本生成任务，通过对输入文本进行改造，构造人工模板，随之对改造后的输入文本进行关键词提取，并将提取后的关键词与改造后的输入文本拼接，从而构造新的输入，接着引入关键词提取算法TF-IDF加强生成模型对关键词的关注，在CNN/DM数据集上的实验表明，该模型能够有效提高生成文本摘要的质量，使Rouge-1、Rouge-2、Rouge-L值得到了提高。更多还原

【Abstract】 Aiming at the problem that the existing pre-trained model is difficult to fine-tune and the generative summarization algorithm is prone to generate unregistered words, an abstracting method based on keyword extraction and prompt learning is proposed. First add prompt learning by adding prompt information to the input, change the downstream task into a text generation task, construct an artificial template by transforming the input text, and then perform keyword extraction on the transformed input text, and stitch the extracted keywords with the transformed input text to construct a new input, and then introduce the keyword extraction algorithm TF-IDF to strengthen the attention of the generation model to keywords, and experiments on the CNN/DM dataset show that The model can effectively improve the quality of generated text abstracts, so that the Rouge-1, Rouge-2, and Rouge-L values are improved.更多还原