Social Media Event Prediction using DNN with Feedback Mechanism
Wanlun Ma,Xiangyu Hu,Chao Chen,Sheng Wen,Yang Xiang,Kim-Kwang Raymond Choo
DOI: https://doi.org/10.1145/3522759
2022-03-30
ACM Transactions on Management Information Systems
Abstract:Online social networks (OSNs) are a rich source of information, and the data (including user-generated content) can be mined to facilitate real-world event prediction. However, the dynamic nature of OSNs and the fast-pace nature of social events or hot topics compound the challenge of event prediction. This is a key limitation in many existing approaches. For example, our evaluations of six baseline approaches (i.e., logistic regression latent Dirichlet allocation based logistic regression, multitask learning, long short-term memory and convolutional neural networks, and transformer-based model) on three datasets collected as part of this research (two from Twitter, and one from a news collection site 1 ), reveal that the accuracy of these approaches is between \(50\% \) and \(60\% \) , and they are not capable of utilizing new events in event predictions. Hence, in this paper we develop a novel DNN-based framework (hereafter referred to as event prediction with feedback mechanism – EPFM . Specifically, EPFM makes use of a feedback mechanism based on emerging events detection to improve the performance of event prediction. The feedback mechanism ensembles three outlier detection processes and returns a list of new events. Some of the events will then be chosen by analysts to feed into the fine-tuning process to update the predictive model. To evaluate EPFM, we conduct a series of experiments on the same three datasets, whose findings show that EPFM achieves \(80\% \) accuracy in event detection and outperforms the six baseline approaches.We also validate EPFM’s capability of detecting new events by empirically analyzing the feedback mechanism under different thresholds.