Base

L

% of correct prediction of text fragment sequence

Advantage of NFIM (%)

By unconditional probability

By 1^st fragment

By 2^nd fragment

By mean probability

Method 1

Method 2

Method 3

Method 4

NFIM

“Russian”

1

35.62

24.32

22.82

18.64

18.65

15.56

16.06

17.26

10.87

43.15

2

9.97

6.59

6.54

5.37

5.37

5.09

4.98

4.94

4.13

19.61

“German”

1

21.45

16.25

12.07

11.40

11.86

10.58

10.08

9.88

6.97

41.75

2

11.78

8.25

6.97

6.61

6.62

6.03

6.00

5.66

4.55

24.40

“Ukrainian”

1

35.60

23.27

21.60

20.50

20.58

15.66

16.74

14.11

8.65

63.12

2

1.22

0.72

0.75

0.66

0.66

0.61

0.63

0.57

0.54

5.56

“English”

1

29.85

18.37

17.77

14.17

14.28

12.78

11.71

12.22

8.33

40.58

2

34.04

22.67

22.16

17.55

17.56

16.92

16.13

16.46

13.40

20.37