Research On Grey Prediction Of Network Public Opinion Heat Driven By Data Characteristic

More and more netizens obtain,publish and forward information through Weibo and other social platforms recently,and Internet has increasingly become an important medium for people to express their attitudes.However,online public opinion emergencies occur frequently,which damages the stability of the online environment and brings pressure to relevant departments.Compared with traditional public opinion,online public opinion spreads quickly,which means that online public opinion may experience fermentation,outbreak,recession and other evolutionary trends in just a few hours.Considering the grey characteristics of online public opinion,the grey system theory is applied to the hotness prediction of online public opinion events,and new grey models applicable for online public opinion prediction are proposed in this paper.The specific studies are as follows:(1)In view of the time lag effect existing in the spread of online public opinion,the relationship between system behavior and different variables is analyzed,the dynamic time-delay parameters are introduced to the grey model,and the corresponding delay value is determined according to the grey relational degree among related factor sequences and the system characteristics sequence.In addition,in order to reduce the influence of the randomness of online public opinion data on the prediction results,the fractional order accumulation operator is introduced to establish the fractional order accumulation delay multivariable discrete grey model,which is effective to the short-term prediction of network public opinion.(2)Considering that online public opinion spread quickly and there is no aftereffect in the dissemination of information,the new information variable weight buffer operator is proposed,which follows the principle of “New information priority”.The new information is assigned larger weight under the action of the buffer operator.The new information variable weight fractional order accumulating generated operator is proposed on the basis of the buffer operator,which has the dual functions of accumulation and weighting.In addition,the idea of metabolism is introduced to the model to establish the new information variable-weight fractional rolling grey model.The model is not only applicable to the prediction of monotone sequence,but also applied to the prediction of oscillation trend of online public opinion(3)In view of the characteristics of seasonal fluctuation of online public opinion,an improved seasonal grey decomposition and ensemble model is proposed in this paper.The STL decomposition algorithm is used to decompose original public opinion data.And the grey modified exponential model is proposed based on the grey difference information.Then the dynamic seasonal factors are introduced to establish the seasonal modified exponential grey Bernoulli model.The new model is used to predict the seasonal sequence and trend sequence,and the ARIMA model is used to predict the remainder sequence.(4)In view of the nonlinearity of online public opinion trend,a damping accumulated multivariable grey model is proposed to forecast online public opinion trends in this paper.Firstly,the dynamic damping trend factor is introduced into the accumulation process,so that the model can adjust the accumulating order of different sequences more flexibly.Secondly,considering that the accumulated sequences have grey exponential rate property,the damping grey multivariable model is established by optimizing the structure of background values.Finally,due to the assumption that the relevant factor variables are grey constants,the systematic error occurs in the traditional grey multivariate model,the time response equation is given to reduce error by using the composite quadrature method.