基于知识图谱和语义相似度的岗位匹配与排序方法  

Position matching and ranking method based on knowledge graph and semantic similarity

作　　者：何春辉 郭博譞 HE Chunhui;GUO Boxuan(School of Mathematics and Computational Sciences,Xiangtan University,Xiangtan,Hunan 411105,China;Beijing Dublin International College,Beijing University of Technology,Beijing 100124,China)

机构地区：[1]湘潭大学数学与计算科学学院,湖南湘潭411105 [2]北京工业大学北京-都柏林国际学院,北京100124

出　　处：《湖南城市学院学报（自然科学版）》2021年第5期59-63,共5页Journal of Hunan City University:Natural Science

摘　　要：互联网的普及极大地促进了在线招聘平台的发展,如何利用有效的算法在海量的职位和简历库中自动匹配符合岗位要求的简历,是构建在线招聘系统的难点之一。为解决上述问题,本文融合知识图谱和文本语义相似度算法,提出了一种采用字符搜索寻找符合岗位要求的简历子集和根据文本语义相似度对所得子集进行排序的两阶段方法。实验结果表明,基于BERT和余弦的语义相似度排序方法的平均准确率比基线方法高15.3%,即可显著提高岗位与简历的匹配度。The popularity of the Internet has greatly promoted the development of online recruitment platforms.How to use effective algorithms to automatically match the resumes that meet job requirements in a large number of positions and resume database is one of the difficulties in building an online recruitment system.In order to solve the above problems,this paper integrates the knowledge graph and the text semantic similarity algorithm,and proposes a two-stage method to find the resume subset that meets the job requirements through character search and sorts the subsets according to the text semantic similarity.The experimental results show that the average accuracy of the semantic similarity ranking method based on BERT and cosine is 15.3%higher than the baseline method,which can significantly improve the matching degree of the position and the resume.

关 键 词：知识图谱 语义相似度 排序 命名实体识别 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]