Skip to main content

Table 5 Crosslingual experiments (zero shot)

From: Advances in monolingual and crosslingual automatic disability annotation in Spanish

 

System

Precision

Recall

F-measure

 

A−

A+

A−

A+

A−

A+ (stdev)

 

\(FLAIR_{MUSE}\)

70.51

76.11

23.72

27.94

35.43

42.58 (±4.40)

Train ENG

\(FLAIR_{ME}\)

54.65

61.24

28.67

37.70

37.31

46.31 (±2.31)

Test SPA

\(FLAIR_{mBERT}\)

35.04

38.85

10.04

12.66

13.98

17.12 (±9.67)

 

\(XLM-R\)

48.09

58.04

2.57

28.61

29.49

37.51 (±9.35)

 

\(FLAIR_{MUSE}\)

57.08

59.41

23.32

35.01

32.89

43.71 (±2.87)

Train SPA

\(FLAIR_{ME}\)

64.71

69.95

28.40

41.84

39.45

52.34 (±1.44)

Test ENG

\(FLAIR_{mBERT}\)

17.53

22.70

13.03

19.76

14.49

20.45 (±13.21)

 

\(XLM-R\)

18.93

28.26

6.16

10.49

8.71

14.28 (±4.82)

  1. Results of the different approaches used for automatic disability annotation in Spanish trained on English data. A−: without Acronym and abbreviation module. A+: with the Acronym and abbreviation module