Important event trend prediction method based on HC-TC-LDA

The invention discloses an important event trend prediction method based on HC-TC-LDA. The important event trend prediction method includes the steps: firstly, adopting a web crawler to crawl specialnews data; then, extracting features through a topic model, and designing a topic model HC-TC-LDA fusing hierarchical clustering, word contribution degree and LDA for the problems that an LDA topic model has a large number of noise words and the number of topics needs to be manually set according to experience; and finally, constructing an important event trend prediction model by adopting a machine learning classification model. The prediction performance of the important event trend prediction method is superior to that of a traditional method which depends on expert knowledge to construct characteristic indexes. The important event trend prediction method can effectively realize high-precision prediction of the trend of the important event, has certain advantages in universality and timeliness, does not need mo