多模态内容资源编目及知识表示研究

Research on Cataloging and Knowledge Representation of Multimodal Resource

在多模态信息资源体量快速增长的环境下,为更有效地对内容资源开展编目及知识表示工作,本文从多模态的角度出发,探讨在“实体-属性-关系”结构下多模态资源编目可行性,提出一种适用于多模态资源的编目框架M-RDA、梳理内容资源模态归一化方法。本文重点扩展编目框架中的多模态属性与关系类目,从4个模态维度出发,挖掘多种资源间的描述补充关系,在著录资源的同时对其模态特征及关系进行标注。此外,为实现编目规则下知识的组织和表达过程,本文从事件描述的原因、对象、地点、时间、人员、方法6个方面提出一种基于5W1H的多模态知识表示方法,实现多模态资源从标签获取、编目到知识表示的完整过程,为多模态资源的编目自动化提供理论参考和实践案例。

In order to carry out the work of cataloging coding and knowledge representation for content resources effectively under the background of rapid growth of multi-modal information resources,this paper discusses the feasibility of multimodal resource cataloging under the structure of“entity-attributerelationship”,based on the RDA,and proposes a cataloging framework M-RDA.This paper reorganizes the method of content resource modal normalization.This paper focuses on expanding the multimodal attributes and relationship categories in the cataloging framework.It explores the description and supplementary relationship between multiple resources from four modal dimensions,and annotates their modal characteristics and relationships while cataloging resources.In addition,in order to realize the organization and expression process of knowledge under the cataloging rules,this paper proposes a multimodal knowledge representation method based on 5W1H to describe events from six aspects:cause,object,place,time,person and method,which completes the process from tag acquisition,cataloging and knowledge representation,providing a theoretical basis for the automation of multimodal resource cataloging.