Performance of named entity recognition using SVM learn-to-rank (SVM+LTR) for all entities in the cross-validation test and unique entities only.

Each horizontal row shows Precision, Recall and F-score performance for a class using alternative methods. Unique Rate shows the percentage of unique entity mentions seen in the cross-validation test for each class. ALL shows micro-averaged F-score.每行展示了采用不同方法时，单个类别的精确率（Precision）、召回率（Recall）与F分数（F-score）的性能表现。唯一提及率（Unique Rate）用于表征各分类在交叉验证测试中出现的唯一实体提及（entity mentions）所占百分比。ALL项展示微平均F分数（micro-averaged F-score）。