大语言模型在学科知识图谱自动化构建上的应用  

Application of Large Language Models in Automated Construction of Knowledge Graphs for University Subject Domains

作　　者：唐晓晟[1] 程琳雅 张春红[1] 朱新宁[1] TANG Xiaosheng;CHENG Linya;ZHANG Chunhong;ZHU Xinning(School of Information and Communication Engineering,Beijing University of Posts and Telecommunications,Beijing 100876,China)

机构地区：[1]北京邮电大学信息与通信工程学院,北京100876

出　　处：《北京邮电大学学报（社会科学版）》2024年第1期125-136,共12页Journal of Beijing University of Posts and Telecommunications(Social Sciences Edition)

基　　金：北京邮电大学2022年教育教学改革项目(2022JXYJ-F01);2022北京市高等教育本科生教学改革与创新项目。

摘　　要：人工智能技术的飞速发展推动了教育领域的智能化,涌现出一些利用知识图谱技术进行学科知识体系构建的研究。利用知识图谱构建的知识体系,可以建模知识点之间的关联性,形成课程知识脉络,有助于学习者对知识点的记忆与深层次理解。然而,学科课程的教学资源多样且分散,现有的自动构建方法往往使用单一的数据资源,资源利用率低,难以对专业知识体系的构建提供有效指导。构建方法自动化程度较低,构建成本大,需要开发不同的模块对非结构化文本进行处理并生成图谱,可复现性和可移植性较差。因此,提出了一种基于大语言模型的学科知识图谱构建优化流程,高度融合大语言模型的优势和各学科知识的逻辑关联。具体以通信专业为例,借用ChatGPT大语言模型的强大功能,使用多种知识来源,通过学科知识分析设计知识系统本体,自动化提取基于教学资源的知识实体和知识关系形成最终的学科知识体系,可方便地推广至其他学科领域。The rapid development of artificial intelligence has promoted the intellectualization in the field of education,and research on the construction of subject knowledge system by using knowledge graphs has emerged.The knowledge system constructed by knowledge graphs can model the correlation among knowledge points and form the course knowledge context,which is helpful for learners to memorize and understand knowledge in a deep level.However,the teaching resources of subject courses are diverse and scattered,and the existing automatic construction methods often use a single data resource with low resource utilization rate,which is difficult to provide guidance for the construction of professional knowledge systems.Moreover,low degree of automation and high cost need different modules to process the unstructured texts and generate the graphs,so reproducibility and portability are weak.Therefore,an optimization process of subject knowledge graph construction is proposed based on the large language model,which highly integrates the advantages of the large language model and the logical correlation of knowledge in various subjects.Specifically,taking the major of communication as an example,the powerful function of a large language model such as ChatGPT is used to design a knowledge system through subject knowledge analysis by using a variety of knowledge sources,so as to automatically extract knowledge entities and knowledge relations based on teaching resources to form a final subject knowledge system,which can also be introduced to other subject fields.

关 键 词：大语言模型 知识图谱 智能化教育 ChatGPT 自动化图谱构建 

分 类 号：G434[文化科学—教育学]