Table 5 Confusion matrix for multi-class cyber hate classification using a combined dataset of 3 different protected characteristics
From: Us and them: identifying cyber hate on Twitter across multiple protected characteristics
a | b | c | d | e | f | ← classified as |
|---|---|---|---|---|---|---|
1,798 | 3 | 61 | 0 | 1 | 0 | a = non-hate-disability |
18 | 31 | 2 | 0 | 0 | 0 | b = hate-disability |
74 | 0 | 1,724 | 7 | 0 | 1 | c = non-hate-race |
0 | 0 | 28 | 42 | 0 | 0 | d = hate-race |
3 | 0 | 3 | 0 | 1,577 | 37 | e = non-hate-sexual orientation |
0 | 0 | 0 | 0 | 108 | 75 | f = hate-sexual orientation |