Research On History Based Employment Recommendation And Visual Analsis

With the development of society and the successive growth of the number of college graduates, the employment of college graduates is more and more getting the attention of government, schools and students. Employment recommendation is a process of recommending suitable positions to students, in which the critical problem is how to select the appropriate recommendation algorithm and how to efficiently find the accurate "nearest neighbors". At present, there are some relatively perfect platforms for Job-hunting. However, when recommending jobs to the user, these platforms only consider user’s information and recruitment conditions of company, but ignore the history employment information of the company and don’t analyze the students employed by the company in previous years. For the above reasons, the recommendation result is defective.Aiming at the problems existing in the existing employment recommendation systems, in this thesis we put forward an employment recommendation algorithm based on history information and asymmetric sierpinski carpet to solve the problems above. In this thesis, the main idea is:first access to the employment data and filter the data, analyze the employment information and model by comprehensively using a variety of methods, then we can calculate the similarity between different companies and produce recommendation results. Moreover, when demonstrating the result, we adopt the technique of sierpinski carpet to implement demonstrating in layer.The employment recommendation algorithm based on history information and asymmetric sierpinski carpet put forward in this thesis mainly involves three parts, including employment recommendation algorithm, visualization techniques of recommendation result and user interaction technology. The research in this thesis has important meaning in three aspects. First, save the valuable time and effort of users and companies by recommending appropriate positions to users through analyzing users’information and companies’employment history information. Second, improve the accuracy of employment recommendation by comprehensively using a variety of methods to analyze the employment information from various aspects. Third, accelerate users’ employment process and enhance cognitive ability. In the process of demonstrating the result, we use visualization technology and interactive technology to describe the distinction among the results, which is helpful for users to find deeper relationship and do the employment decisions.Aiming at the problems of ignoring employment history information existing in the existing employment recommendation systems, the algorithm put forward in this thesis fully takes the employment history information into account, whose main idea is to forecast the recruitment trend through the employment history. The employment history information of one company has great reference value to the students’ employment choice; moreover, we can analyze and calculate the similarity between different companies through the recruitment history information. When calculating the similarity, we should adopt different models to calculate quantifiable factors and unquantifiable factors relatively.In order to demonstrate the recommendation result, we adopt the data visualization technique based on sierpinski carpet. The existing sierpinski carpet is symmetric and all the pieces divided in the same step are all in the same size, which results in that the distinction between different pieces can’t be demonstrated. In order to solve the problem above, data visualization based on asymmetric sierpinski carpet is put forward, from which we can demonstrate the recommend results with different importance in layer.