Research Of Recommendation System For The Peer-To-Peer Lending Products Based On Hybrid Algorithms

In recent years,with the advent of the Internet Finance,In C hina,the P2P(Peer-To-Peer)lending platform growth has explode.For investors who faced with so many P2 P lending platforms are hard to quickly find the lending products they need,the results can lead to a large number of potential users and excellent lending products just miss the person or opportunity.In view of this,hybrid recommendation method is proposed in this paper.the specific content of the recommendation method for P2 P lending products is as follows.(1)Improved k-means clustering algorithms.K-means clustering algorithm was improved with particle swarm optimization algorithm to determine the initial center point,and AUTO-K algorithm optimization category K to determine the value.(2)Improved based on bipartite graph structure of recommendation algor ithm.One is simply to improve the edge weight by the amount(WNBI).Another is adding time factor of user interest shift model(SNBI).(3)Hybrid recommendation.Weighted linear mixed WNBI model and SNBI model are SWNBI,the experimental results show that this model is more effective in accuracy.According to the interpretation of clustering results,we selected a list of products for new users.Giving priority to recommend a top-N product list according to the model of SWNBI and randomly selected top-N product correspondence cluster category in the list of products for users who have history records.The innovation of this paper is as follows.(1)Improved K-means clustering algorithm is combined with the particle swarm optimization algorithm and AUTO-K algorithm,and the improved clustering method is used twice in this paper.(2)Considering two aspects of the time factor and the amount of investment factors to improve the traditional NBI model.(3)The recommendation list according to the product list of randomly selected clustering results could solve the problem of long tail platform and products.