摘要
针对档案信息利用成本高、利用率低、编研困难等问题,本研究实现了基于深度学习的档案多模态智能编纂方法。考虑到编纂知识的准确性和多样性,结合NLP及其图像处理技术将档案知识进行了关键信息抽取;为了保证生成内容与实际业务相符且较为规范,设计了档案业务主题模板及其编纂规则,并结合Chat GLM实现了档案主题内容智能编纂。
In view of the problems of high utilization cost,low utilization rate and difficult compilation and research of archival information,a multi-modal intelligent compilation method based on deep learning is studied and implemented.Considering the accuracy and diversity of compilation knowledge,the key information of archival knowledge is extracted by combining NLP and its image processing technology.In order to ensure that the generated content is consistent with the actual business and relatively standardized,the archival business topic template and its compilation rules are designed,and the intelligent compilation of archives theme content is realized by combining ChatGLM.
作者
刘伊玲
王胡燕
王聪杰
杨本富
Liu Yiling;Wang Huyan;Wang Congjie;Yang Benfu
出处
《兰台世界》
2024年第7期79-85,共7页
Lantai World
基金
云南电网有限责任公司信息中心—面向档案海量数据关键信息提取和智能编纂原型研究应用(项目编号:YNKJXM20220102)。
关键词
多模态关键信息抽取
档案智能编纂
智能生成
深度学习
multi-modal key information extraction
intelligent compilation of archives
intelligent generation
deep learning
