Table 2 Cross validation of different types of cyber hate
From: Us and them: identifying cyber hate on Twitter across multiple protected characteristics
Training data | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
Disability | Race | Sexual orientation | ||||||||
P | R | F | P | R | F | P | R | F | ||
Testing Data | Disability | 0.96 FP = 1 | 0.61 FN = 20 | 0.75 | 0.00 FP = 0 | 0.00 FN = 51 | 0.00 | 0.00 FP = 1 | 0.00 FN = 51 | 0.00 |
Race | 0.00 FP = 1 | 0.00 FN = 70 | 0.00 | 0.87 FP = 7 | 0.64 FN = 25 | 0.74 | 0.95 FP = 1 | 0.29 FN = 50 | 0.44 | |
Sexual orientation | 0.00 FP = 2 | 0.00 FN = 183 | 0.00 | 1.00 FP = 0 | 0.09 FN = 165 | 0.18 | 0.74 FP = 23 | 0.37 FN = 116 | 0.49 | |