Research And Implementation Of P2P Lending Agency's Public Opinion Situation Judgement Based On Sentiment Time Series

In today's information age,the development of hot events spreads timely and rapidly under the influence of the powerful Internet.Online social media,such as Weibo in China,has played an essential role in the process of spreading public opinions and events.The development and changes of public opinions can be effectively reflected by emotional analysis of social network texts.At the same time,the prediction and judgment of public opinion development can also play a better role in assisting decision-making and effective management.Therefore,sentiment analysis for hot events in online social media texts and judgment of public opinion development have become popular topics in recent years.At present,research on textual sentiment analysis mainly aims at single text,and using sentimental features to judge sentiment aspect of the text.However,there is a general lack of such studies on integrated analysis of multi-user and multi-document in unit time for time series.Moreover,most of the existing methods are focused on the information extracted from text itself,including traditional textual features such as sentimental words and part-of-speech.However,research on the domain of text and the feature of scene association is inadequate,such as the feature of identity differences,time sequence of different users and texts on social platforms.Hence,this thesis works on the public opinion texts about the specific events of P2P lending agencies on social network platforms like Weibo,and combines this textual information with sentiment time series,then a text representation including sentimental features and temporal information is proposed to achieve multi-document sentiment prediction.After considering related features of different social user identities and time series,a time+user dual attention mechanism model is developed to analyze and predict the textual information of P2P lending agency 's public opinion.The effectiveness of the proposed model is verified through experiments based on real data from a popular Chinese microblog platform called Sina Weibo.The results show that the values of F1 are 0.837,0.756 and 0.850,respectively,in different data sets,which are higher than other five deep learning algorithms such as LSTM.At the same time,in order to scientifically and quantitatively reflect the current dynamics and future tendency of public opinion,a public opinion index system for P2P is constructed,and the heat trend of public opinion is calculated.By comparing the experiment results with real data,the results show that the accuracy rate is more than 98%and the characteristics of public opinion at different stages are effectively depicted.On this basis,this thesis combines all these research work and implements a prototype system for public opinion situation judgement.