基于潜在语义索引的职位描述聚类研究

A Study of Job Description Clustering Based on Latent Semantics Index

本研究鉴于某公司职位描述存在的特点以及其所带来的问题,对其展开研究。由于职位描述中的文本字数少、数量多,特征维数较高,样本特征稀疏,不能很好的抽取出文本关键特征。针对这些问题,我们用潜在语义索引模型(LSI)对文本进行特征提取,分析潜在语文空间维度对聚类性能的影响,然后根据提取的特征进行K-means聚类,能有效降低简历匹配筛选过程中的职位类别数,提高了简历匹配的效率。

We start this research,in the view of the characteristics of the job description of a company and the problems it brings. Because the job descriptions have the characteristics of less text word,large quantity,sparse sample characteristics and high dimension,the text key characterstics cannot be extracted well. Thus,we use the latent semantic index（LSI） model for feature extraction to analyze the influence of latent semantic spatial dimensions on clustering performance,and the features of extraction are clustered with K-means algorithm,which effectively reduces the number of the job category in the process of resume screening and improve the efficiency of matching the resume.