Skip to main content

Table 2 Results for various LSTM-RNNs and word embeddings on the GM and JNLPBA data sets

From: Long short-term memory RNN for biomedical named entity recognition

Systems

Dim.

GM (P/R/F1 score)

JNLPBA (P/R/F1 score)

LSTM-RNN

   

  +SENNA

50

83.87/80.46/82.13

67.50/72.52/69.92

  +Biomedical

50

85.85/84.09/84.96

70.69/74.80/72.69

  +Google

300

83.90/82.80/83.35

69.19/72.56/70.83

  +Biomedical

300

86.66/85.58/86.12

70.34/74.96/72.58

  +Random

300

83.63/76.56/79.94

66.96/71.46/69.13

BLSTM-RNN

   

  +SENNA

50

84.29/79.83/82.00

67.00/71.60/69.22

  +Biomedical

50

88.42/82.63/85.43

71.04/74.45/72.71

  +Google

300

85.02/82.04/83.50

68.59/73.99/71.19

  +Biomedical

300

87.85/85.29/86.55

71.24/76.53/73.79

  +Random

300

82.87/77.65/80.18

68.43/70.98/69.68