Skip to main content

Table 2 The distribution of pairs for each corpus according to classification success level using cross-learning setting

From: A detailed error analysis of 13 kernel methods for protein-protein interaction extraction

 

AIMed

BioInfer

HPRD50

IEPA

LLL

 

Total

T

F

T, %

F, %

Total

T

F

T, %

F, %

Total

T

F

T, %

F, %

Total

T

F

T, %

F, %

Total

T

F

T, %

F, %

0

41

0

41

0.0%

0.8%

319

319

0

12.6%

0.0%

1

0

1

0.0%

0.4%

9

9

0

2.7%

0.0%

3

3

0

1.8%

0.0%

1

73

6

67

0.6%

1.4%

362

362

0

14.3%

0.0%

4

2

2

1.2%

0.7%

19

17

2

5.1%

0.4%

5

4

1

2.4%

0.6%

2

199

26

173

2.6%

3.6%

322

312

10

12.3%

0.1%

7

3

4

1.8%

1.5%

33

32

1

9.6%

0.2%

10

9

1

5.5%

0.6%

3

315

39

276

3.9%

5.7%

303

280

23

11.0%

0.3%

23

10

13

6.1%

4.8%

38

36

2

10.7%

0.4%

19

19

0

11.6%

0.0%

4

489

71

418

7.1%

8.6%

321

260

61

10.3%

0.9%

27

15

12

9.2%

4.4%

48

45

3

13.4%

0.6%

25

25

0

15.2%

0.0%

5

606

84

522

8.4%

10.8%

355

239

116

9.4%

1.6%

27

15

12

9.2%

4.4%

44

32

12

9.6%

2.5%

25

20

5

12.2%

3.0%

6

547

94

453

9.4%

9.4%

400

208

192

8.2%

2.7%

41

22

19

13.5%

7.0%

51

34

17

10.1%

3.5%

26

18

8

11.0%

4.8%

7

725

136

589

13.6%

12.2%

432

190

242

7.5%

3.4%

43

18

25

11.0%

9.3%

63

32

31

9.6%

6.4%

20

7

13

4.3%

7.8%

8

721

132

589

13.2%

12.2%

586

146

440

5.8%

6.2%

52

17

35

10.4%

13.0%

69

35

34

10.4%

7.1%

34

18

16

11.0%

9.6%

9

767

110

657

11.0%

13.6%

737

95

642

3.7%

9.0%

61

18

43

11.0%

15.9%

107

36

71

10.7%

14.7%

34

19

15

11.6%

9.0%

10

574

118

456

11.8%

9.4%

1060

79

981

3.1%

13.8%

50

14

36

8.6%

13.3%

110

13

97

3.9%

20.1%

56

8

48

4.9%

28.9%

11

414

69

345

6.9%

7.1%

1906

29

1877

1.1%

26.3%

52

16

36

9.8%

13.3%

131

6

125

1.8%

25.9%

50

12

38

7.3%

22.9%

12

363

115

248

11.5%

5.1%

2563

15

2548

0.6%

35.7%

45

13

32

8.0%

11.9%

95

8

87

2.4%

18.0%

23

2

21

1.2%

12.7%

  1. The distribution of pairs (total, positive and negative) in terms of the number of kernels that classify them correctly. Results shown for each corpus separately. Aggregated results are shown in Figure 2. All but the PT kernel are considered. (PT is extremely slow and provide below average results).