Human Resource Recommendation Based On Recurrent Neural Network

In recent years,the employment problem is an urgent social problem.It need be solved.With the rapid development of network technology,the cost of online recruitment is low and human resource recruitment is gradually networked.Corporate recruiters spend a lot of time and energy sifting through a huge pool of resumes to find the right candidate.Facing different job demands,Recruiters not only need to judge whether the job seeker’s academic qualifications and other information meet the recruitment requirements,but also need to analyze whether the job seeker’s work experience and professional skills meet the requirements and description of the position.Screening is very time-consuming and laborious.Therefore,this article proses a personalized resume recommendation program to reduce the difficulty of screening and improve the efficiency of screening.The key idea is to make use of abundant job application data and corporate invitation data.First,we use deep learning models to process free text in resumes and positions,and then extract semantic entities from the entire resume and positions.By fusing free text features and entity features,a comprehensive description of resumes and job positions can be obtained.Specifically,we propose a neural network based on BiLSTM and CNN to extract word-level features of job requirements and resume text,and use an attention mechanism to measure the importance of recruitment requirements to different work experiences and the contribution of work experience to different recruitment requirements.The main research contents include:1.Realizing data collection and preprocessing,Structured processing of job applicant information,recruitment information and other data.Performing data cleaning,Data extraction and Data conversion on the collected data.2.According to the characteristics of job applicant’s resume and recruitment information,this article divides it into structured text and unstructured text.Different texts use different matching methods,Structured text uses numerical matching,Unstructured text is based on a combination of text content matching and semantic similarity matching.Finally,the matching degree ranking is finally obtained by weighting and summing the values of different dimensions.3.In this article,we propose a cyclic convolutional neural network model to process complex text information of resumes and texts.There is a sequence relationship between the resume text and recruitment information,and BiLSTM is used to extract the features in the text.The text content varies in length.In order to better extract the interactive information in the text,an attention mechanism is used to capture features.CNN performs very well in capturing key information and phrase information,so CNN is used to capture local features in the text.The combination of the three can effectively match job requirements and resume text information.