P2P Platform Risk Assessment Model Based On Public Opinion Analysis And Text Theme

In recent years,P2P(peer to peer)lending platform scale has been a rapid increase.At the same time,the bankruptcy situation of the lending platform is also very serious.So how to assess the risk of the P2P platform is the first priority.There are few studies on the risk assessment model of online loan platform,and the public opinion analysis and text information of platform are not considered.There are also differences in evaluation standard.In this regard,the aim of this dissertation is to establish a scientific and effective risk assessment model for P2P platform(LapSVM-PT).To solve above problems,we propose the solutions as follows.First,public opinion on platform are analyzed,based on word segmentation and document vector model on the platform of the comments and news,to get public opinion in different characteristic vector dimension;Then we use theme model on the platform introduction and annual report,to get text topic distribution in different dimensions;Then we choose science the index information of P2P platform,get the index vector model;Finally,we use the semi supervised support vector machine for all the information of platform for experiment,get the final risk assessment results of each P2P platform.In the experimental stage,this dissertation combined with JZT and the WDZJ data by way of experiment.Firstly,we train the dimension of public opinion feature vector and text topic distribution.And then analyze the influence weight of public opinion and text information at the same time.Finally we compare the optimization model and other existing models,the experimental results show that the model in of P2P platform risk assessment field has significant advantages than existing others.