A prediction method of microblog forwarding volume based on ts-lstm and dnn

Abstract

The invention belongs to the technical field of network information processing, and discloses a microblog reposting volume prediction method based on TS-LSTM and DNN, using all microblogs posted by users as source data, extracting 10 user features as input, microblog The magnitude of forwarding is used as the output, and the TS-LSTM prediction model is established to predict the forwarding index of the specified microblog; according to the user's own influence, the influence of the user's fans, the similarity of the microblog features between the user and the fans, the characteristics of the microblog and the prediction The retweeted specified Weibo’s retweet index characteristics are used as input, and whether active fans will retweet is used as output. Finally, the number of retweets will be counted to obtain the final retweet level of the predicted target Weibo. The TS-LSTM algorithm proposed by the present invention is combined with DNN to model and predict microblog forwarding, and the experimental demonstration has at least 5% improvement compared with the previous method.

Description

technical field [0001] The invention belongs to the technical field of network information processing, and in particular relates to a microblog forwarding volume prediction method based on TS-LSTM and DNN. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: [0003] Due to its convenient application and open information platform, Weibo has become an important platform for information diffusion and public opinion dissemination. Microblog forwarding is an important way of microblog dissemination, and it is also one of the most critical issues in the study of microblog information dissemination and public opinion monitoring. Solving this problem can not only better understand the characteristics of public opinion dissemination on Weibo, and lay a foundation for public opinion analysis and monitoring; it can also recommend Weibo or advertisements to users by mining the interest characteristics of users’ Weibo content...

AI Technical Summary

Problems solved by technology

[0007] (1) In the calculation of user influence, PageRank is simply used to calculate the number of fans and followers of the user, without combining more user characteristics, so the calculation of influence is not accurate

[0008] (2) At the same time, with the development of the Internet, the number of Weibo users increases, and the time spent by users on Weibo increases, and the amount of reposts of Weibo will also increase with time. Previous literature did not take this issue into consideration

[0009] (3) For the prediction of Weibo forwarding volume, most of the prediction models are established based on the user’s Weibo characteristics and historical Weibo forwarding volume, and the Weibo forwarding volume is not connected with the user’s fans, which will inevitably lead to the final inaccurate result

Images