Design And Implementation Of Job Recommendation System Based On Resume Data

A resume is a brief description of an individual's life experience.Most of it is used for job search screening.In addition,many of the information in the resume is not fully utilized.However,the personal job search experience can often be used to refer to other similar job seekers.Therefore,this article hopes to find suitable job recommendation for job seekers through in-depth exploration of resume information,including resume project experience and personal information,and hope to help other job seekers through their job search experience.In view of the inefficient treatment methods in the current job market,this article recommends the position by mining resumes to improve efficiency and effectiveness.The first problem is that many companies' resume selection relies on manual or simple keyword matching of academic qualifications,gender,professional,and working years,and then manually screens them again,resulting in a long process of recruitment and low efficiency.Secondly,For job seekers,the current job search website recommendation mostly relies on the keyword matching of the job seeker's intention position.The number of jobrecommendations is large,but it is not necessarily suitable for the job seeker.It takes a lot of time and effort to identify,so based on the above situation,this book The recommendation system will recommend a suitable and more secure job listing to the job seeker based on the resume information uploaded by the job seeker,and adjust the recommendation result according to the position preference of the clicked job to update the job list and display it on the personal center of the job search website.The contributions of article have three way:1)Use the Hadoop-based big data platform to build user models for users.Convert specific fields from the user's resume into numeric fields and store the data in the user library.At the same time,using the deep learning method of fast Text,training the job classification model,and getting the industry label corresponding to the user,the efficiency of the whole recommendation system has been greatly improved.2)Using content-based and demographic-based hybrid recommendation models to find similar users for users,the mixed recommendation model combines the advantages of the two algorithms,including the judgment of the user's personal attributes.Consideration of the project and past job search experience.Determine the impact factor of the impact algorithm,use the analytic hierarchy process to obtain its corresponding weight,and then perform local weighting to obtain similar users.3)Use the Flume log collection system to aggregate the logs into the kafaka cluster.Kafaka then sends the data to Sparkstreaming for real-time processing and also stores the datain HDFS for offline analysis.Sparkstreaming gets a list of popular jobs based on log data,and users can click on the job list to update or get the latest job recommendation list via web page update.At the same time,Flume also stores the data in the offline data disk.The system periodically processes the offline data on the disk,obtains the user's browsing record table,and adjusts the recommendation list according to the user's click browsing condition to make it more in line with the user's needs.Finally,phased system functional testing and performance testing of the entire system.The system research and development is basically completed,and it has not been online yet.The historical log-feeding log data is used to perform the offline recommendation effect test of the system.The experimental results prove that the system can basically meet the job-requirement requirements recommended by the user's suitable position,and can be used as a tool to improve enterprises and job seekers.The efficiency also adds a recommended exploration for the job recommendation field.