【作者】 王静帆； 夏云庆； 郑方； 邬晓钧；

【Author】 Jingfan Wang1,Yunqing Xia2,Fang Zheng2,Xiaojun Wu3 (1. Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China; 2. Center for Speech and Language Technologies, Research Institute of Information Technology, Tsinghua University, Beijing 100084, China; 3. National Laboratory for Information Science and Technology, Tsinghua University, Beijing 100084, China)

【机构】 清华大学计算机系； 清华大学信息技术研究院语音和语言技术中心； 清华信息科学与技术国家实验室；

【摘要】 招聘信息检索与传统信息检索存在较大差异,传统检索方法不能实现良好的招聘信息检索效果。为解决该问题,本文提出二阶段招聘信息检索方法,针对招聘信息的标题文本和职位描述文本分两阶段分别进行不同的处理。第一阶段本文采用VSM模型对标题文本进行初步检索,将相关度较高的招聘信息视为种子;第二阶段,本文采用文本相似度度量方法和聚类分析方法,在招聘信息全集中寻找与种子相似度较高的招聘信息。通过结合"请求-文档"相关度和"文档-文档"相似度,最终计算相似招聘信息与搜索请求的相关度,完成检索结果综合排序。实验证明,这个方法能有效提高招聘信息检索系统的性能。更多还原

【Abstract】 Job information retrieval holds unique characteristics compared to traditional information retrieval leading to lower performance. In this paper, text structure of the job posting is analyzed and a two-step framework is proposed to address this problem. In the first step, the traditional vector space model (VSM) model is applied on the title field to get relevant job post-ings. In the second step, the top relevant documents are considered as seeds to retrieve similar job posting using document similarity measures and clustering technique on full text. Finally, relevant scores between the similar job postings and the query are calculated by combing relevant scores in the first step and similarity scores in the second step. Our experiment proves that this method is effective in job information retrieval.更多还原