Multimodal Information Social Media Popularity Prediction Method Based on Iterative Optimization Strategy

Abstract

The invention discloses a multi-modal information social media popularity prediction method based on an iterative optimization strategy, and aims to overcome the defect of insufficient utilization ofmulti-modal data and extract multi-modal features from posts. In order to enhance the aging stability of the features, post features are averaged in a sliding window, then multi-modal feature fusion is performed, and regression prediction is performed on the fused features by using a LightGBM model. In order to solve the difficulty of popularity extremum prediction, an iterative optimization strategy is provided, and the residual error of the predicted popularity score, especially extremum compensation, is effectively compensated. Through a large number of experiments carried out on an SMPD2020 data set, a good effect is achieved, and the effectiveness of the method is proved.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a multi-modal information social media popularity prediction method based on an iterative optimization strategy. Background technique [0002] With the development of Internet technology and the rise of smart terminal devices, social media has become an important part of people's lives. Different from traditional media, modern social media platforms such as Flickr, Facebook, and Twitter rely more on user relationship networks for information exchange and dissemination. However, due to the information overload in the network and the limited attention of users, information on social media has uneven attention. Therefore, using user information, posts, etc. to predict social media popularity has high research value and commercial value, and can also help content creators to produce more popular works. [0003] The existing social media popularity prediction work is...

AI Technical Summary

Problems solved by technology

In the part of social media feature extraction, most of the current methods are based on single-modal text data, ignoring the image and user feature data and the correlation between the popularity of different posts of the same user. Inadequate utilization often leads to one-sidedness and unsatisfactory forecast results

In terms of model regression prediction, most of the current machine learning-based popularity prediction methods use smooth regularization terms to avoid overfitting, but this regression method will lead to smoothing of the prediction results, that is, the extreme value of the popularity is difficult to predict. Accurate prediction, however in real life, posts with greater popularity tend to have greater value, and it is very important to correctly predict these extreme values

Images