Research On Recruitment Feature Recommendation System Based On User Profile

With the continuous development of the Internet,many online recruitment platforms have been established in China.The advancement of the Internet has brought many changes to recruitment.The emerging online recruitment platforms facilitate companies and job seekers to view recruitment information,but there are also some problems.The massive increase in information has led to information overload.Recruitment information is buried in the massive amount of information,and users cannot obtain the effective information that they really want;information asymmetry and ineffective matching of information also affect the recommendation results of the recruitment platform.This paper proposes a research on the feature recommendation system of online recruitment information based on user portraits in view of the existing recruitment platform problems.First,this article discusses the theory and application of user portraits and recommendation systems in the context of big data,and describes the status quo,problems,and characteristics of recruitment information of existing online recruitment platforms.A method of using NLP text processing is proposed for the large amount of text data appearing in job applicants and recruitment information,that is,using the combination of TF-IDF and SVD algorithms to extract the features in the job applicant’s resume and job details and reduce the dimensionality to construct user portrait features set.Secondly,the user portrait feature data set obtained by text processing is input into the Light GBM model and the feature recommendation algorithm based on Naive Bayes,and the feature data set is trained to obtain the feature importance ranking,position recommendation sequence and recommendation results.Finally,the experimental results are evaluated through the evaluation indicators such as accuracy,precision,and predicted matching value commonly used in recommendation algorithms.This paper uses the text feature extraction method to extract the text features of job applicants and recruitment information to get better feature information.Then through the Light GBM model and the feature recommendation algorithm based on Naive Bayes,the acquired job applicants and recruitment data feature sets are trained on feature data sets,and finally better recommendation results and accuracy rates are obtained compared with other model methods.It can be seen that the use of better user portrait feature extraction methods and machine learning model training allows the online recruitment platform to better adapt to realistic recommendation scenarios.