基于关键词和Transformer的文本摘要生成研究

Research on Text Summarization Generation Based on Key Words and Transformer

生成技术在生成摘要时忽略了关键词的作用,导致生成的摘要难以聚焦关键信息。为此,提出一种以Transformer模型为基础结构,融合关键词和卷积神经网络的文本摘要生成方法,实现以关键词为引导的摘要生成。实验在CSDS数据集上进行,结果表明该方法在ROUGE指标上均有提升,验证了其有效性。

Traditional summarization generation techniques overlook the role of keyword information in generating summaries,resulting in difficulty in focusing on key information in the generated summaries.To solve this problem,a text summarization generation method based on the Transformer model,and integrating keywords and convolutional neural networks is proposed to achieve keyword guided summarization generation.The experiment is conducted on the CSDS dataset,and the results shows that the method improved on the ROUGE scores,verifying the effectiveness of the proposed method.