Table 6 MRR and Hit@1 scores (percentage) with different history lengths m on ICEWS14 and GDELT datasets. The best result for different history lengths is highlighted in black font.

From: A contrastive learning framework with dual gates and noise awareness for temporal knowledge graph reasoning

History length (m)

1

2

3

4

5

6

7

8

9

10

ICEWS14 (MRR)

50.27

50.63

50.70

50.64

50.53

50.39

51.18

50.57

50.37

50.25

ICEWS14 (H@1)

39.23

39.70

39.72

39.74

39.86

39.19

40.37

39.59

39.22

39.15

GDELT (MRR)

22.98

23.59

23.00

22.37

23.35

21.98

23.27

23.07

20.26

21.68

GDELT (H@1)

13.84

14.78

13.73

13.22

14.17

13.00

14.07

13.92

11.32

12.32

Back to article page