Predicting Emerging Trends of Keywords based on Graph Neural Network

摘要

In order to achieve better prediction results on emerging trends of keywords, a graph neural network is used to mine the relationship between keywords. This paper first constructs a keyword network based on co-occurrence relationship. Then, a feature template is carefully designed. Finally, the relationships among keywords are captured by a graph neural network, and the method for predicting emerging trends is proposed. The results on the test set show that the average Euclidean distance of the prediction results of this method is 12.33, and the rank correlation is 0.85, which significantly outperforms the baseline methods.