在当今大数据的时代下，各类信息数据呈几何级增长，也冲击着企业的生存与发展，如 何利用信息技术手段改善当前工作方式，提高企业运行效率至关重要。企业之间竞争的本质 就是人才的竞争，人才的质量与数量决定公司发展能否长久。随着信息技术越来越发达，互 联网已基本普及的背景下，互联网招聘已经成为国内大多数企业日常最常用的人才招聘方式， 面对互联网上的海量人才简历，如何快速准确地发现并录用符合公司发展需要的人才，成为 近几年众多企业和专家学者们共同关注的焦点。 简历筛选工作作为人力资源招聘工作重要的一个环节，它涉及到后续企业面试、录用等 环节，此环节筛选的候选名单质量也决定企业人才的质量。当前企业在简历筛选方面更多是 靠人为经验以及主观性的判断开展，这对简历筛选人员的能力素质要求较高。当前人力资源 的评估针对岗位要求和筛选的要求，往往通过长期工作经验积累有一套固定的筛选办法，而 实际却是公司各个岗位人才要求结合公司当前发展阶段的需要是动态调整的。再者，面对当 前互联网上各种第三方招聘平台的简历样式与格式的不一致问题，大大增加了人力筛选的难 度，效率的低下也就制约了公司的发展。 本文以人力资源中招聘环节的成本为出发点，从减少人为简历筛选效率低下问题。具体 从简历筛选前岗位人才模型构建、互联网平台简历关键数据获取，通过机器学习算法相结合 作为简历自动筛选的方案。本文在参考了前人研究的基础上，提出基于企业人才要求变化而 动态调整的简历筛选方法，有效降低了企业人力招聘成本，提高了人才招聘效率。 本文结合大数据IT公司南京诚勤教育科技有限公司，以实际数据研发工程师岗位招聘为 例来实际验证提出的互联网简历筛选方案。通过两组样本来进行验证，第一组为100份简历 样本，第二组在包含100份简历样本的基础上再增加100份简历样本，通过机器进行学习分 别推算得出符合条件的筛选名单。第一轮验证在不改变方案情况，观察在数据样本增大的情 况下，原有样本的子集得出的符合名单是否一致来验证互联网简历筛选方案的稳定性；第二 轮验证在调整了相关参数后，方案运行得出简历筛选匹配率是否针对性变化。通过样本的测 试发现，本方案具备一定的实用价值，也可以为后续其他企业在互联网简历筛选工作中提供 支撑。 关键词: 招聘，简历筛选，人才模型， 机器学习 II Abstract In today's era of big data, all kinds of information data show geometric growth, but also impact the survival and development of enterprises, how to use information technology means to improve the current way of work, improve the operation efficiency of enterprises is crucial. The essence of competition among enterprises is the competition of talents. The quality and quantity of talents determine whether the company can develop for a long time.With more and more developed, information technology has been largely under the background of popularization of Internet, Internet recruitment has become the most domestic enterprises daily the most common way of recruitment, faced with the enormous talent resume on the Internet, how to rapidly and accurately found that meets the needs of company development and talent, become experts in recent years, many enterprises and scholars common focus. Resume screening, as an important part of human resource recruitment, involves the basis of subsequent interview and employment, and the quality of candidate list also determines the quality of talents in the enterprise.At present, enterprises are more dependent on human experience and subjective judgment in resume screening, which will have higher requirements on resume screening personnel.The current evaluation of human resources is aimed at the requirements of job requirements and screening, often through long-term work experience accumulated a set of fixed screening methods, but in fact it is the company's various positions of talent requirements combined with the needs of the company's current development stage is dynamically adjusted.In addition, faced with the inconsistency of resume style and format of various third-party recruitment platforms on the Internet, the difficulty of manpower selection is greatly increased, and the low efficiency also restricts the development of the company. This paper starts from the cost of recruitment in human resources to reduce the inefficiency of artificial resume screening.Specifically, it will combine the construction of job talent model before resume screening and the acquisition of key resume data on the Internet platform with machine learning algorithm as the program of resume automatic screening.On the basis of previous studies, this paper proposes a resume screening method based on dynamic adjustment of enterprise talent requirements, which, to some extent, reduces the cost of human resources recruitment and improves the efficiency of talent recruitment. In this paper, nanjing chengqin education technology co., LTD., a big data IT company, took the recruitment of real data r&d engineers as an example to verify the proposed Internet resume screening scheme. Two sets of samples were used for verification. The first group was 100 resume samples, and the second group was 100 resume samples on the basis of 100 resume samples. The qualified screening list was calculated by machine learning.In the first round, without changing the scheme, the stability of Internet resume screening scheme was verified by observing whether the conforming list obtained from the original sample subset was consistent with the case that the data sample was enlarged.In the second round of verification, after adjusting the relevant parameters, the scheme runs to determine whether the matching rate of resume screening is targeted.Through the sample test, it is found that this scheme has certain practical value, and can also provide support for other enterprises in the Internet resume screening work.. Keywords: recruitment, resume screening, talent model, machine learning III 目录 第一章 绪论 .. 1 1.1 研究背景 ........................ 1 1.2 研究的意义 .................... 2 1.2.1 理论意义 ............. 2 1.2.2 实践意义 ............. 2 1.3 研究思路和技术路线 .... 3 1.3.1 研究思路 ............. 3 1.3.2 研究的技术路线 . 3 1.4 研究的主要内容 ............ 3 1.5 可能的创新之处 ............ 4 第二章 研究的理论综述 ............. 5 2.1 人力资源招聘研究综述 5 2.1.1 传统的招聘方式 . 5 2.1.2 互联网招聘方式 . 5 2.1.3 初创企业的招聘 . 6 2.2 简历筛选研究概述 ........ 6 2.2.1 简历筛选的意义 . 6 2.2.2 简历筛选的评价 . 7 2.2.3 简历筛选的方法 . 8 2.3 机器学习研究综述 ........ 9 第三章 诚勤公司招聘中的简历筛选现状及问题分析 .......................... 12 3.1 诚勤公司简述 .............. 12 3.1.1 公司简介 ........... 12 3.1.2 诚勤公司的人力资源招聘现状....... 12 3.1.3 诚勤公司人力资源招聘中的简历筛选现状 .................. 13 3.2 诚勤公司互联网简历筛选工作访谈 ......... 14 3.2.1 访谈对象 ........... 14 3.2.2 对行政人员的访谈设计及综述....... 15 3.2.3 对公司管理者的访谈设计及综述... 16 3.2.4 对员工的访谈设计及综述 .............. 16 3.2.5 对外部专家的访谈设计及综述....... 17 3.3 诚勤公司简历筛选工作问题总结分析 ..... 18 3.3.1 简历筛选依赖招聘人员专业性....... 18 3.3.2 公司人才模型缺失 .......................... 18 3.3.3 简历匹配方法没有建立 .................. 19 3.3.4 简历筛选的质量缺少追踪 .............. 19 第四章 诚勤公司基于机器学习的简历筛选模式... 21 4.1 简历筛选前的准备工作 ............................. 21 4.2 人才模型构建方法 ...... 22 4.2.1 人才评价模型的设计原则 .............. 22 4.2.2 人才评价指标选择方法 .................. 23 4.2.3 权重分类及分配方法 ...................... 24 4.2.4 小结 ................... 24 4.3 简历筛选人才指标设定 ............................. 26 IV 4.3.1 能力方面 ........... 27 4.3.2 态度方面 ........... 28 4.3.3 荣誉方面 ........... 29 4.4 简历数据采集 .............. 31 4.4.1 制定简历采集策略 .......................... 31 4.4.2 简历采集策略执行 .......................... 35 4.5 机器学习算法筛选 ...... 38 4.5.1 系统聚类算法 ... 39 4.5.2 K-means聚类算法 ............................ 40 4.6 简历筛选结果 .............. 41 4.7 本章小结 ...................... 43 第五章 诚勤公司简历筛选的保障与优化措施....... 45 5.1 科学制定招聘计划 ...... 45 5.1.1 梳理公司岗位需求 .......................... 45 5.1.2 选择招聘渠道 ... 46 5.1.3 核验招聘结果 ... 46 5.2 公司人才模型构建常态化 ......................... 46 5.3 简历有效性复核机制 .. 47 5.3.1 机器筛选会存在分类标签的不完整性 .........