“中文+专业”知识图谱的构建及应用前景——以“中文+土木工程”为例  

Construction and Application Prospects of the“Chinese+Major”Knowledge Graph:The Case of“Chinese+Civil Engineering”

作　　者：肖锐 侯尚余 张邝弋 XIAO Rui;HOU Shangyu;ZHANG Kuangyi(School of International Chinese Language Education,Yunnan University,Kunming 650091,China)

机构地区：[1]云南大学汉语国际教育学院,云南昆明650091

出　　处：《云南师范大学学报(对外汉语教学与研究版)》2025年第1期63-74,共12页Journal of Yunnan Normal University(Teaching & Studying Chinese as a Foreign Language Edition)

基　　金：国家语委“十四五”科研规划2022年度重点项目“中文+职业技能”教学资源建设研究(ZDI145-35);云南省教育厅科学研究基金资助项目、云南大学第四届专业学位研究生实践创新基金项目“面向南亚东南亚的‘中文+工程’多模态知识图谱构建与应用研究”(ZC-242410029)。

摘　　要：当前“中文+专业”教育需求迅猛增长,如何高效地利用中文组织、检索并应用专业内的教学资源,已成为实现个性化精准教学亟待解决的挑战。以“中文+土木工程”为例,借助知识图谱对专业内容进行知识组织与可视化展示,能有效提升信息检索效率、促进跨领域知识的理解与应用。研究提出新的知识图谱构建路径:首先,分别基于LDA模型、BERT-LDA模型以及BERTopic模型对“中文+土木工程”文本内容进行主题建模;其次,基于Silhouette主题轮廓指数计算不同主题建模方法的聚类质量,以确定主题聚类最优的方案;然后,通过依存句法分析对“中文+土木工程”文本聚类结果进行语义三元组抽取,从而构建了“中文+土木工程”文本知识模型,共形成11大类的29549组实体以及31470组关系,为搭建适应“中文+土木工程”领域特点的知识图谱语义架构奠定了坚实基础;最后,利用Neo4j图数据库及本体编辑工具对“中文+土木工程”的知识本体进行构建与存储,并用Cypher查询语言进行知识建模。经理论构型与实际验证,本研究完整构建了基于主题聚类的“中文+土木工程”知识图谱,不仅为相关领域的研究人员提供了丰富的数据资源和强大的知识检索,而且该技术路径还具备良好的迁移性,对推动“中文+专业”交叉学科的发展具有重要意义。The demand for“Chinese+Major”education is currently experiencing rapid growth.How to efficiently use Chinese to organize,retrieve and apply professional teaching resources has become a challenge that awaits solution to achieve personalized and precise teaching.Taking“Chinese+Civil Engineering”as an example and using knowledge graphs to organize and visualize the subject-specific content can effectively improve information retrieval efficiency and promote the understanding and application of cross-domain knowledge.This study proposes a construction approach for the new knowledge graph:first,the“Chinese+Civil Engineering”text content is modeled based on the LDA model,BERT-LDA model and BERTopic model;secondly,the clustering quality of different topic-modeling methods is calculated based on the Silhouette topic-based index profile to determine the optimal topic-clustering scheme;then,the semantic triples of the“Chinese+Civil Engineering”text-clustering results are extracted through syntactic dependency parsing,thereby constructing the“Chinese+Civil Engineering”text knowledge model,forming a total of 29,549 entities and 31,470 relationships in 11 categories,laying a solid foundation for building a semantic architecture of the knowledge graph that adapts to the characteristics of the“Chinese+Civil Engineering”program;finally,the Neo4j graph database and the ontology editing tool are used to construct and store the“Chinese+Civil Engineering”knowledge ontology,and the Cypher query language is used for knowledge modeling.After theoretical configuration and practical verification,this study has fully constructed a“Chinese+Civil Engineering”knowledge graph based on thematic clustering.It not only provides rich data resources and powerful knowledge retrieval for researchers in related fields but also demonstrates good transferability,signifying its importance in promoting the interdisciplinary development of“Chinese+Major”.

关 键 词：中文+专业 中文+土木工程 主题建模 知识图谱 教学资源 

分 类 号：H195[语言文字—汉语]