Skip to main content

Table 8 Type-based learn to rank on training data

From: Concept selection for phenotypes and diseases using learn to rank

ID

Sys

P

R

F1

ID

Sys

P

R

F1

T019

SVMRank

0.46

0.15

0.23

T047

SVMRank

0.53

0.64

0.58

 

ListNet

0.47

0.13

0.20

 

ListNet

0.42

0.65

0.51

 

RankNet

0.47

0.11

0.18

 

RankNet

0.43

0.57

0.49

 

RankBoost

0.43

0.16

0.23

 

RankBoost

0.46

0.68

0.55

T020

SVMRank

0.32

0.57

0.41

T048

SVMRank

0.44

0.62

0.52

 

ListNet

0.29

0.50

0.36

 

ListNet

0.42

0.54

0.47

 

RankNet

0.33

0.42

0.37

 

RankNet

0.39

0.51

0.44

 

RankBoost

0.21

0.50

0.30

 

RankBoost

0.48

0.67

0.56

T033

SVMRank

0.01

0.32

0.02

T184

SVMRank

0.46

0.63

0.53

 

ListNet

0.01

0.48

0.02

 

ListNet

0.40

0.66

0.50

 

RankNet

0.01

0.48

0.03

 

RankNet

0.39

0.62

0.48

 

RankBoost

0.01

0.45

0.02

 

RankBoost

0.45

0.64

0.53

T037

SVMRank

0.34

0.33

0.33

T190

SVMRank

0.20

0.61

0.30

 

ListNet

0.26

0.24

0.25

 

ListNet

0.17

0.58

0.27

 

RankNet

0.22

0.21

0.22

 

RankNet

0.15

0.44

0.22

 

RankBoost

0.30

0.29

0.29

 

RankBoost

0.16

0.60

0.25

T046

SVMRank

0.29

0.66

0.40

T191

SVMRank

0.31

0.56

0.40

 

ListNet

0.25

0.67

0.36

 

ListNet

0.37

0.44

0.40

 

RankNet

0.27

0.61

0.37

 

RankNet

0.35

0.36

0.36

 

RankBoost

0.24

0.67

0.35

 

RankBoost

0.35

0.56

0.43

  1. Note that the highest scoring system F1 for each semantic type is shown in bold.