Research And Implementation Of Enterprise-oriented Policy Push Algorithm

Policy push algorithm is a kind of recommendation algorithm applied in the field of government services,and its related research has been landed in recent years to promote the implementation of policies and ensure that enterprises can enjoy policy dividends in a timely manner.The current policy push algorithm can basically meet the needs of enterprises to quickly match with policies,but the accuracy of recommendation is suboptimal,especially in data sparse scenarios.To address the above problems,this paper proposes an enterprise-oriented policy push algorithm,which mines the implicit similarity factors among enterprises and uses the policy data features and the neighbor relationship among enterprises to iteratively predict the enterprises’ attitudes toward potential policies in order to improve the recommendation accuracy.To address the problem that the policy push algorithm lacks the comprehensive utilization of implicit factors such as feature information,extreme attitude and common rating proportion when calculating the similarity between enterprises,this paper proposes a similarity algorithm based on multiple types of implicit factors.The algorithm calculates similarity from two dimensions of rating information and feature information,and introduces extreme rating similarity and common rating coefficient in the rating dimension to make full use of the implicit similarity information between enterprises and improve the accuracy of similarity relationship between users.To improve the recommendation accuracy of the policy push algorithm in sparse data scenarios,this paper proposes an improved iterative rating prediction algorithm incorporating features.Aiming at the personalized differences in the rating matrix,this paper proposes a mapping method between ratings and satisfaction.In the prediction process,the algorithm uses the policy data features and the neighbor relationship between enterprises to iteratively predict the attitude of unknown items for the target users.The algorithm introduces a reliability matrix to correct the influence of the distance between neighbors and target users on the prediction results,and combines the policy features to improve the accuracy of the prediction performance.Based on the proposed algorithm,this paper designs and implements an enterprise-oriented policy push system.Starting from the policy pushing target,this paper analyzes the system requirements and divides the system into data acquisition and pre-processing layer,data analysis layer and data presentation layer.By summarizing the workflow of each layer,the modules involved are designed and implemented in detail,and test cases are designed respectively to verify the availability of the system.