Data Mining In The Re-employment Application Management

As information technology is widely applied to labor market and social security management, a large amount of data concerning employment, unemployment, job-hunting as well as recruitment, has been accumulated in social security system of various regions. These data are numerous and incomplete, while at the same time they are very useful. Data mining could analyze and sort out these data systematically and effectively, providing users with useful and timely information they need.The thesis aims to investigate the issue of how to apply data mining to re-employment management. It firstly introduces"the Re-employment Management System of XX City"which successfully manages data about unemployed people, enterprises along with recruitment for workers and so forth; by means of ART algorithm and EM algorithm, the system conducts analysis about data from which useful information has be singled out. ART algorithm, a kind of regression algorithm targeting successive value prediction, breaks down the data into major trend elements and time-variation elements. Through the application of ART algorithm into the analysis of the time sequence data of recruitment in XX City from 2006 to 2009, the trends of labor demand in certain industries can be predicted. This prediction model could help these industries understand the law of demand regarding labor resources, which would facilitate the provision of specific vocational skills training, the addressing of differences between labor supply and demand as well as the optimizing of labor resource distribution. EM clustering algorithm is a popular iterative refinement algorithm which can be regarded as an extension of K-means algorithm. Based on the probability of the occurrence of subordination between objects and clusters, the objects can be allotted by this algorithm. 2753 samples were randomly selected from the data of unemployed people registered from 2007 to 2008 in XX City. As to these samples, EM algorithm is used for clustering and distinct features of different categories are also summed up. With reference to these features, specific polices in relation to re-employment facilitation will be formulated, which would help unemployed people for re-employment with well-defined objectives. Through the analysis of data in this thesis, the significance of data mining in re-employment management is persuasively illustrated. Moreover, aided by the prediction and findings obtained via data mining, data support could be provided for labor and social security department to formulate re-employment facilitation policies.