基于知识图谱的语义增强信息检索  

Semantically-enhanced Information Retrieval using Knowledge Graph

作　　者：朱星图 邓远飞 张永华 黄光健 ZHU Xing-tu;DENG Yuan-fei;ZHANG Yong-hua;HUANG Guang-jian(School of Economics and Statistic,Guangzhou University,Guangzhou 510006,China;School of Artificial Intelligence,Guangdong Open University,Guangzhou 510091,China;School of Information Engineering,Jiangxi University of Science and Technology,Ganzhou 341099,China)

机构地区：[1]广州大学经济与统计学院,广东广州510006 [2]广东开放大学人工智能学院,广东广州510091 [3]江西理工大学信息工程学院,江西赣州341099

出　　处：《电脑与电信》2024年第11期55-60,共6页Computer & Telecommunication

基　　金：2024年广东省教育科学规划课题(高等教育专项)(编号:2024GXJK140);江西省教育厅科学技术研究项目(编号:204201400810);江西省高层次人才科研专项项目(编号:205200100500)资助。

摘　　要：近年来,为了能够考虑查询语句和文档间的语义关系,基于语义的检索方法被提出,但许多研究者在构造语义检索方法的过程中存在着语义信息考虑单一、人为主观影响大等问题,针对这些问题提出了一种基于知识图谱扩建语义网的信息检索方法SIRKG。首先结合Jiang等人在语义检索方面的研究成果,利用多个公共知识图谱DBpedia、WordNet等,提出一种新的语义网络AWDSN,从而实现了语义检索。最后,在公用数据集CACM和CISI中与经典检索方法Lucene以及语义检索方法SIRWWO进行了实验对比,证明本文提出的SIRKG方法具有更高的准确度。In recent years,semantic-based retrieval methods have been proposed to consider the semantic relationship between query statements and documents.However,many researchers have encountered problems such as single consideration of semantic information and significant subjective influence during the construction of semantic retrieval methods.Therefore,this paper proposes an information retrieval method SIRKG based on knowledge graph expansion of the semantic web to address these issues.This article first combines the research results of Jiang et al.in semantic retrieval,using multiple public knowledge graphs such as DBpedia and WordNet,proposes a new semantic network AWDSN to achieve semantic retrieval.Finally,this article conducts experimental comparisons with the classical retrieval method Lucene and semantic retrieval method SIRWWO on the common datasets CACM and CISI,demonstrating that the proposed SIRKG method has higher accuracy.

关 键 词：信息检索 知识图谱 语义相关度 多知识源 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]