面向多模态医疗健康数据的知识组织模式研究

Research on the Knowledge Organization Model of Multimodal Medical and Health Data

[目的/意义]随着网络化和数字化的快速发展,线上线下医疗健康活动产生了海量多模态数据,当前迫切需要一种更为完善的知识组织方法,对类型繁多、专业性强的医疗健康多模态数据进行序化和组织。[方法/过程]首先对多模态数据进行预处理以构建多模态医疗健康数据集;接着基于实体对齐和关系抽取实现医疗健康数据的知识单元抽取,并通过知识评估和融合构建多模态知识单元;然后在多模态知识单元和知识图谱基础上提出一种面向多模态医疗健康数据的知识组织模式;最后以文本和图像模态数据为例,以医疗健康问答系统为实践应用进行分析和验证。[结果/结论]针对多模态医疗健康数据的内在特性,基于多模态知识图谱和语义知识组织框架,提出一种面向多模态医疗健康数据的知识组织模式。本研究推进了多模态知识组织的理论深度,可为医疗健康领域知识序化、精准知识服务和深度知识发现提供有效支撑。

[Purpose/Significance]With the rapid development of networking and digitalization,online and offline medical and health activities have generated massive multimodal data.At present,a more perfect method of knowledge organization is urgently needed to sequence and organize medical and health multimodal data with various types and strong expertise.[Method/Process]First,the paper preprocessed the multimodal data to construct a multimodal medical and health dataset;Next,implemented knowledge unit extraction of medical and health data based on entity alignment and relationship extraction,and constructed multimodal knowledge units through knowledge evaluation and fusion;Then,a knowledge organization model for multimodal medical and health data was proposed based on multimodal knowledge units and knowledge graph;Finally,text and image modal data were taken as examples,and the medical health question answering system was used as the practical application for analysis and verification.[Result/Conclusion]In light of the inherent characteristics of multimodal medical and health data,the study proposes a knowledge organization pattern for multimodal medical and health data based on multimodal knowledge graph and semantic knowledge organization framework.This research advances the theoretical depth of multimodal knowledge organization and provides effective support for knowledge serialization,precision knowledge services,and deep knowledge discovery in the field of medical and health care.