| abstract |
The present invention provides a hot topic early development trend prediction system and prediction method, collecting topic time series; judging whether the sequence enters the recession period; if it enters, as a complete topic time series, the clustering method is used to classify each topic class, and the Each topic time series of each topic class is substituted into the prediction model for training to obtain the intra-class prediction model of each topic class; if not entered, analyze the similarity between the new topic time series and each complete topic time series in the topic class, and take the average as the new topic The degree of matching with the topic class; filter out the set number of topic classes according to the order of matching degree from high to low; call the intra-class prediction model of the screened out topic classes, and substitute them into the new topic time series to obtain the predicted value of the set number; The predicted value is assigned to different weight values for combination to obtain the predicted value of the new topic in the future. The above system and method can accurately predict the early development trend of new topics. |