Ranking Of The Achievements Of Common Service Agent Granted By Innovation Fund For Technology Based Firms

After the innovation fund for technology based firms has implemented from 2006,almost 30000 service agencies have been funded.But it is very difficult to evaluate the effectiveness of fund implementation because there are too many attributes.This evaluation now is done by expert,wich is hard to ensure the fairness and causes the heavy workload.Thus we need to research methods which evaluate the effectiveness of the fund by the computer,to overcome the shortcomings of artificial rating method.This paper is organized as follows:Firstly,the paper introduces the research background,significance of the grant projects and current research achivements,respectively,then overviews the performance of rating algorithms and their ineffectiveness.Secondly,the paper overviews the classification algorithms in the field of machine learning.Since the problem of performance rating can be transformed into a multiclassification problem,the collected data of the implementation of the service agency grants fund can be treated as the training set,wich can be used to train a kind of multiclassifier to achieve the rating of the new service agencies grant funds,therefore,we review the current classification algorithm.Thirdly,the attribute reduction approch is proposed based on the fruit fly optimization,and applied to reduct the attribute of the achievements of the subsidy fund.The simulation results show the effectiveness of the method.Finally,the ensemble of multi-class classifiers based on the version space analytic center is used to grade the achievements of the subsidy fund.The simulation results show the effectiveness of the method.