| Affected by the adverse global economic environment,China’s economic growth has slowed down,domestic and foreign demands have fallen,but the number of college graduates has continued to increase year by year,which triggers fiercer employment competition.The current traditional employment mode has some prominent problems of mismatch in supply and demand as well as asymmetric information between recruitment companies and job applicants.On the one hand,job hunters are unclear about recruiters’ employment information and their abilities,which leads to an inefficient employment mode.On the other hand,the recruiters could hardly find the candidates they need after frequent interviews,which results in brain drain and the increase of time cost sometimes.The outbreak of COVID-19 poses huge challenges for college graduates.With the suspension of offline recruitment activities,a great number of graduates are much hindered from seeking for a job.However,online recruitment platforms have many problems such as information overload,lack of targeted recommendation,and neglect of users’ data privacy and safety.Motivated by the issues above,this paper proposes an HR bidirectional recommendation system targeting college graduate recruitment based on data mining and privacy-preserving technologies(HBRS).By analyzing the data of recruiters and applicants,the system utilizes data mining and privacy protection technologies for two-way recommendation.Specifically,the main works and contributions of the paper are as follows: A hierarchical anonymous HR data processing scheme based on entropy weighting method is proposed.The purpose of this scheme is to process the collected HR data through the following two modules,so that the processed data can be used for more effective calculation of HBRS,and to solve different levels of privacy data security problems through hierarchical anonymous model.Specifically,for the entropy weighting method module,an entropy weighting method-based data standardization and vectorization method are proposed firstly.Then we introduce the cluster result evaluation index to evaluate the data.The results show that the data after undergoing treatment via entropy weighting method are superior to those without treatment in terms of clustering analysis.For the hierarchical anonymity module,firstly a kind of anonymous union cascade protection model applicable to data release is proposed.The model is an anonymous model that can choose safety protection grade.Then,we conduct experiments to evaluate and analyze the proposed models,and the significance of adding privacy protection degree is verified.A scheme for HBRS with clustering algorithm based on attribute-weighted similarity is proposed.The purpose of this scheme is to realize more reasonable mining and analysis of HR data through the following two modules.Specifically,for data mining module,Based on the improved K-means centroid selection algorithm(ICSA),we firstly propose a data clustering model,which replaces the steps of choosing centroid in an initialized and random manner mainly by improved algorithm to avoid local optimum.Secondly,we put forward the Data Similarity Calculation Model(IWSA)Based on Improved Weighted Similarity Algorithm,the algorithm calculates the attribute weights by using the entropy evaluation method and introduces weight coefficient in similarity calculation.Finally,through three experiments,the clustering effects of Elbow method,Silhouette Coefficient method and DB index method are compared and analyzed,which proves that DB index method is more suitable for HBRS data.The k value of optimal clustering quantity is determined then via DB index convergence.For recommended algorithm module,the module contains two parts,an employment-oriented recommendation model for content-based recommendation algorithms(EO-CB)and a recruiter-oriented recommendation model for Item-based collaborative filtering recommendation algorithm(RO-ItemCF).The design implements an HR bidirectional recommendation system targeting at the college graduate recruitment.The system contains three interfaces,student login end,enterprise login end,and big data display end.Specifically,Student login end: this end provides colleges and universities with accurate recommendation services.Enterprise login end: the recruiters to screen the advantageous resources of graduates at school effectively use this end.Big data display end: this end achieves data visualization with which the employment office of universities and relevant government organs can monitor the employment situation,make scientific decisions and take measures as soon as possible.It has a broad range of applications.Finally,we also introduce two indexes,recommendation accuracy rate P and recommendation recall ranking index F to certify the effectiveness of the HBRS. |
