Research Of College Student Employment Recommending System Based On Improved K-means Clustering

With the downward pressure on China’s socialist market economy, the international economic situation downturn, the continuous of college expansion, graduates employment has become a serious social problem. Although the main channel for graduate employment and key sources of information is campus recruiting employment information publishing platform, but this approach still has many issue. On the one hand, the contents and manners of job guide of employment information platform provided by the university is single. It primarily regularly publishs recruitment information content and promote the national employment policy for students.It lacks the employment recommended function. On the other hand, graduates need to filter from the huge recruitment information for their own information. The procedure is costly job costs, including time cost, energy cost and opportunity cost.This paper designs and implements the recommendation system for university graduates in the above problems. It is designed to offer graduates a one-to-one reliable employment recommended guidelines, while reducing the blindness of student job search process. System is based on real historical data for students and companies. Factors to consider the individualized student job search process,the system uses recommendation algorithms, data mining, business assessment index model and recommended index weights to get the results. There are three major pieces. (1) The first section is data extraction. It extracts student data and enterprise data from existing information management systems and campus student employment information platform. To enhance the objective system accuracy, using Aprior algorithms filter out herein related issue, delete irrelevant attributes. (2) The second section is students and businesses similarity calculation. By using the improved K-Means clustering algorithm, combined with the similarity SimRank algorithm and ultimately get the similarity between students and businesses. (3) The third section is recommended ranking. Combined with "the similarity between students and business" and "business assessment index model" calculated the rankings for each student recommended. While adding personality filtering capabilities to further improve the hit rate is recommended.The previous student database data is the historical student data from 2009 to 2012 of Dalian Maritime University. Extract 1003 student data in 2012 and part of 2013 as graduating students database to test the system. Extract Maritime University Employment Network 1222 corporate data. There are several key indicators in the evaluation system:recommended number (N), the system is recommended percentage (P), the accuracy of the recommended system(F).System testing and parameters determined by the final recommendation system recommended number of 25, the system hit rate of 0.67, Recommended Rank Index System is 5.9. Recommended model through systematic assessment designed to test the system in line to meet the needs of student employment personalized recommendation.