Table 3

Contribution of linguistic features. Results from various combinations of types of linguistic features, as described in Features section, combined using Vector Space Model learning algorithm. LC = Local Collocations, SB = Salient Bigrams and U = Unigrams.

Features

Data sets

LC

SB

U

SB+U

LC+SB

LC+U

All words

79.2

82.0

86.9

85.9

86.3

86.9

Joshi subset

72.6

74.4

81.6

82.3

81.0

82.0

Leroy subset

66.2

66.9

76.7

77.5

76.5

77.3

Liu subset

75.7

76.2

83.4

84.3

82.7

83.9

Common subset

69.6

77.7

79.3

79.1

77.6

78.8

Stevenson et al. BMC Bioinformatics 2008 9(Suppl 11):S7   doi:10.1186/1471-2105-9-S11-S7