Skip to main content

Table 6 The performance of mention detection with different models on two datasets

From: Distinguished representation of identical mentions in bio-entity coreference resolution

BioNLP

P

R

F1

Bioe2e

82.0

66.3

73.3

BioNeu

84.1

73.1

78.2

BioNeu + SFA

83.4

76.1

79.6

KE-LSTM

78.0

84.1

80.9

KE-LSTM + SFA

78.2

84.3

81.1

CRAFT

P

R

F1

E2E_MetaMap

67.1

52.7

59.0

BERTfilter

73.1

57.8

64.5

BioNeu

81.6

49.6

61.7

BioNeu + SFA

83.4

49.4

62.0

KE-LSTM

79.3

63.1

70.3

KE-LSTM + SFA

78.8

65.6

71.5

  1. E2E_MetaMap and BERTfilter are the baselines in [7]. Bioe2e is the baseline in [9], and KE-LSTM is the baseline in [10]
  2. The maximum value is in bold