Research And Implementation Of P2P Platform Risk Prediction Based On Internet Texts

With the rapid development of the Internet,the financial industry has developed a new business model through the Internet-Peer to Peer lending(P2P lending).P2P lending facilitates financing and lending by small and medium-sized enterprises,while also providing new avenues for personal investment.The P2P platform refers to the online loan platform that operates the P2P lending business.While the P2P platform suffers from difficulties in payment or the ability to repay,the senior management of the company has problems such as greed or money-carrying,and has not been timely managed and supervised,it will generate financial risks,which may trigger financial emergencies.This paper studies and implements P2P platform risk prediction based on Internet text data.Firstly,this paper obtains P2P platform information and P2P platform comment text from P2P third-party information platform through crawler technology,and constructs sentiment classification dataset and sequence labeling dataset in P2P domain by means of manual annotation.Secondly,this paper uses TextCNN model to classify the comments,and the time series of sentiment tendency is obtained,so as to achieve the purpose of measuring the trend of investor sentiment.This paper verifies the relationship between investor's sentiment time series and trading volume index through Granger causality test and Pearson correlation coefficient.Based on the above research,this paper proposes a P2P trading volume prediction method based on investor sentiment changes.It was found that sentimental characteristics played a significant role in the trading volume prediction.Thirdly,this paper proposes an improved sequence labeling model ELMo-BiLSTM-CNN-CRF(EBCC)model and P2P platform risk prediction method.This paper finds that the monthly negative sentiment changes trend of investors is positively correlated with the number of monthly problem platforms,and the investor's sentiment is reflected in the comments of P2P platform.Therefore,the EBCC model is used to extract key information from investor comments which these informations contains the true views of investors on the P2P platform.This paper uses the text features and data features of the comments to train the P2P platform risk prediction model to provide a useful method for preventing financial emergencies.Finally,the paper discusses the specific implementation of data collection,data processing and model.