Skip to main content

Advertisement

Table 2 Performance comparison of different models including Precision/Recall/F1-score

From: Detection of medical text semantic similarity based on convolutional neural network

Model Macro average Positive class Negative class AUC (mean ± std)
Precision (mean ± std) Recall (mean ± std) F1-score (mean ± std) Precision (mean ± std) Recall (mean ± std) F1-score (mean ± std) Precision (mean ± std) Recall (mean ± std) F1-score (mean ± std)
Zero-r 0.73 ± 0.0 0.85 ± 0.0 0.78 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.85 ± 0.0 1.0 ± 0.0 0.92 ± 0.0 0.0 ± 0.0
Keyword Mapping 0.827 ± 0.006 0.842 ± 0.005 0.833 ± 0.005 0.464 ± 0.023 0.358 ± 0.018 0.404 ± 0.018 0.891 ± 0.004 0.927 ± 0.004 0.909 ± 0.003 0.840 ± 0.004
LSA 0.892 ± 0.005 0.862 ± 0.007 0.873 ± 0.006 0.512 ± 0.022 0.758 ± 0.019 0.611 ± 0.019 0.956 ± 0.004 0.879 ± 0.008 0.916 ± 0.005 0.894 ± 0.006
LDA 0.872 ± 0.006 0.852 ± 0.006 0.860 ± 0.006 0.514 ± 0.021 0.669 ± 0.023 0.581 ± 0.019 0.936 ± 0.005 0.884 ± 0.005 0.910 ± 0.004 0.879 ± 0.004
Doc2Vec 0.882 ± 0.007 0.862 ± 0.007 0.869 ± 0.007 0.514 ± 0.019 0.682 ± 0.023 0.586 ± 0.018 0.943 ± 0.005 0.892 ± 0.006 0.917 ± 0.004 0.871 ± 0.005
NER-based 0.835 ± 0.006 0.849 ± 0.005 0.842 ± 0.006 0.473 ± 0.022 0.501 ± 0.020 0.482 ± 0.020 0.904 ± 0.006 0.923 ± 0.005 0.912 ± 0.005 0.853 ± 0.004
Siamese LSTM 0.920 ± 0.006 0.891 ± 0.005 0.904 ± 0.006 0.582 ± 0.020 0.843 ± 0.021 0.698 ± 0.020 0.964 ± 0.006 0.901 ± 0.007 0.932 ± 0.006 0.916 ± 0.006
CNN + random vector 0.916 ± 0.005 0.931 ± 0.006 0.923 ± 0.005 0.631 ± 0.022 0.833 ± 0.019 0.712 ± 0.019 0.972 ± 0.007 0.917 ± 0.005 0.941 ± 0.005 0.942 ± 0.003
CNN + pretrain vector 0.912 ± 0.006 0.927 ± 0.006 0.920 ± 0.006 0.637 ± 0.021 0.811 ± 0.019 0.701 ± 0.020 0.965 ± 0.006 0.920 ± 0.005 0.937 ± 0.006 0.936 ± 0.004
CNN + concept vector 0.931 ± 0.006 0.938 ± 0.007 0.935 ± 0.006 0.682 ± 0.023 0.771 ± 0.020 0.734 ± 0.021 0.969 ± 0.004 0.938 ± 0.008 0.954 ± 0.007 0.951 ± 0.003