Dynamic evolution monitoring of brand sentiment based on causal discovery neural networks

In this paper, we deal with the shortcomings of the old brand sentiment watching ways which are unable to catch those deep cause-effect links as well as their ever-changing development trends by presenting a new kind of brand sentiment watching way -a dynamic growth watching system for brand sentiment using causal finding neural networks.This model can find the causal structure between different sentiment elements by itself with a structured causal discovery module.Combined with a temporal neural network to model the sentiment evolution path, it can do causal inference and dynamic prediction of sentiment trend.From the experiments, we see that our method is much better than the other models on all the metrics, and the improvement is statistically significant.