基于依存关系与主题差异的非遗知识元抽取方法研究  

Research on Intangible Cultural Heritage Knowledge Elements Extraction Based on Dependency Relations and Subject Differences

作　　者：董坤[1] DONG Kun

机构地区：[1]中南民族大学图书馆,湖北武汉430074

出　　处：《图书情报导刊》2023年第1期72-79,共8页Journal of Library and Information Science

基　　金：中南民族大学教研项目“双一流背景下面向教学、研究工作的文献资源保障体系建设研究”(项目编号:JYX20123);湖北省高等学校图书情报工作委员会重点项目“高校图书馆文献资源建设现状及其优化策略研究”(项目编号:2020-ZD-04);中央高校基本科研业务费专项资金项目“基于依存句法分析的民族文献知识元及其语义关系抽取研究”(项目编号:CSY20029)。

摘　　要：为准确抽取出细粒度非遗知识元,实现非遗知识的深度揭示与有效利用,提出了一种基于依存句法关系与主题差异的知识元抽取方法。该方法分为3个阶段:第一阶段根据统计特征识别出非遗术语;第二阶段根据非遗术语及其依存关系实现非遗知识元知识项的半自动抽取;第三阶段基于知识元的主题结构实现对冗余、噪声数据的过滤。该方法在非遗项目“摆手舞”文档集上进行了实验验证,表明了其可行性与有效性。In order to accurately extract the fine-grained knowledge elements(KEs)in the field of intangible cultural heritage and achieve depth disclosure and effective use of intangible cultural heritage knowledge,this paper proposes a method for KEs extracting based on dependency syntactic relations and subject differences,which is divided into three stages:in the first stage the terms of intangible cultural heritage are identified according to the statistical characteristics,in the second stage the semi-automatic extraction of intangible cultural heritage knowledge items of KEs is realized accord⁃ing to the terms and their dependency relations,and in the third stage the subject structure of KEs is used to filter redun⁃dant and noisy knowledge items.The method is tested on the document set of the intangible cultural heritage project“wav⁃ing hands dance”,indicating its feasibility and effectiveness.

关 键 词：非物质文化遗产 依存分析 潜在狄立克雷分布 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]