Research On Universities Personalized Employment Recommendation System Based On Collaborative Filtering

With the increasing number of college graduates in china, the problem of hard employment has become more apparent. However, the students can not grasp the enterprises information effectively. They do not have clear employment target, lack of employment information and good preparation.This may have a negative effect on employment. In the face of massive recruitment information from employment website and the information asymmetry between students and enterprises, it is difficult for graduates to search enterprises which fit for them. Graduates often go to jobs fair blindly.It’s not only waste time and energy, but also miss some suitable jobs and reduce successful rate of employment. Facing the present situation, although all colleges and universities carry out some employment guidance and recommendations, but because of the amout of graduates, it’s difficult to give personalized recommendation to each graduate based on their characteristics. In addition, the current college employment website also can not recommend fit jobs for graduates.It just can release employment information. Therefore, we have to find an objective, personalized and targeted measure imminently.With the research and application of personalized recommendation system, we can use personalized recommendation system to solve the employment recommendation.By mining employment intention, students professional interest, school performance and other aspects of information, combined with the previous employment data, employment recommendation system can recommend suitable employment unit for graduates.In this way, graduates can not only prepare to the job directly, but also reduce time and energy.lt can improve the successful rate of employment. At present, research on the universities personalized employment recommendation system is not established.Recommendation model, algorithm and effect need to be improved. This thesis focus on the three issues which still need to be improved in the current study.(1) Lack of recommendation which combine with students’ characteristics. At present, the traditional collaborative filtering algorithm is widely used. But it just depends on the employment interest score of students, without considering the the impact of students’ employment features.(2) It can not compute the impact weights of the employment features objectively. At present, subjective evaluation method is widely used, but it can not reflect actual situation.(3) Although k-means clustering algorithm can improve the speed of recommendation, however, there is not still resolved the problem that k-means algorithm is affected by the initial clustering center.In this thesis, in order to solve the problem, firstly, the factors which affect the employment of graduates are analyzed and 9 students’ employment characteristics are extracted. Secondly, the information gain ratio is selected as the method to calculate feature weight. Thirdly, in order to avoid the initial cluster centers may affect the clustering results, an AK-means algorithm for clustering of students is proposed. Then MATLAB programming is used to verify the validity of the algorithm. Finally, combined with the employment characteristics of student and interest scores, collaborative filtering employment recommendation model is constructed. Based on the recommendation model, the 4 years students’ employment data of one engineering college is used to verify and analyze the model. The employment recommendation prototype system has been developed based on B/S structure, SQL Sever 2008 database and C# programming language.It can realize the recommendation function.Through the verification, the proposed employment recommendation model has a certain validity.It can provide reference for the students.This thesis has positive significance to explore the application of the recommendation system in colleges and universities.