Summary of accuracy, precision, recall, and F1 scores of the models based on different types of the features.

Models with the tweet features yielded better performance than those with the author features. When all features were used, the models showed the best. Best model for each case is highlighted.

采用推文特征（tweet features）的模型，其性能优于采用作者特征（author features）的模型。当整合使用全部特征时，模型的综合表现达到最佳。各实验场景下的最优模型均已高亮标注。