多模态知识图谱构建及其在智慧知识服务中的应用  

Construction of Multimodal Knowledge Graph and Its Application in Intelligent Knowledge Service

作　　者：闫永君 庄媛 YAN Yong-jun;ZHUANG Yuan(Library of Luxun Academy of Fine Arts,ShenYang 110004,China;Library of Criminal Investigation Police University of China,ShenYang 110854,China)

机构地区：[1]鲁迅美术学院图书馆,沈阳110004 [2]中国刑警学院图书馆,沈阳110854

出　　处：《大学图书情报学刊》2024年第6期112-117,共6页Journal of Academic Library and Information Science

基　　金：2024年辽宁省高等学校图书情报工作委员会项目“生成式人工智能驱动智慧图书馆建设研究”(LTB202420)。

摘　　要：伴随大数据、人工智能、5G等技术的迅速发展,数据量越来越大,数据形态也多种多样,多模态知识图谱能够发现语料文本中的实体知识信息,并建立实体知识关系,形成一个实体知识网络。多模态知识图谱的应用场景丰富,作为文献资源和知识服务中心的图书馆也非常重视对多模态资源的组织利用。文章以教育领域为研究对象,采用深度神经网络识别数据实体,实现对图片、视频等多模态数据的表达,通过基于开放数据融合的知识抽取、基于多特征的跨领域知识融合,构建了教育领域多模态知识图谱,并对其在某高校中的实体匹配效果和课程重构中的应用效果作出分析。以期拓宽多模态知识图谱在图书馆智慧知识服务中的应用思路。With the rapid development of big data,artificial intelligence,5G and other technologies,the amount of data is getting larger and larger,and the forms of data are also diverse.Multi-modal knowledge graph can discover entity knowledge information in corpus text,establish entity knowledge relations,and form an entity knowledge network.The application scenarios of multi-modal knowledge graph are abundant,and libraries,as document resources and knowledge service centers,attach great importance to the organization and utilization of multi-modal resources.Taking the field of education as the research object,this paper adopts deep neural network to identify data entities and realize the expression of multi-modal data such as pictures and videos.Through knowledge extraction based on open data fusion and cross-domain knowledge fusion based on multi-features,a multi-modal knowledge map in the field of education is constructed.It also analyzes the effect of entity matching and curriculum reconstruction in a university to broaden the application of multi-modal knowledge graph in library intelligent knowledge service.

关 键 词：多模态知识图谱 实体识别 关联性 智慧知识服务 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术] G252[自动化与计算机技术—计算机科学与技术] G250.7[文化科学—图书馆学]