Table 6 Statistical values on the split of toxic dialogues by interlocutor for each category predicted by Detoxify.
Toxic Split | Expert | User | ||||||
|---|---|---|---|---|---|---|---|---|
mean | std | min | max | mean | std | min | max | |
toxicity | 2.4E-03 | 2.4E-02 | 3.0E-04 | 9.9E-01 | 3.4E-02 | 1.6E-01 | 3.0E-04 | 2.4E-03 |
severe toxicity | 4.0E-06 | 4.2E-05 | 1.0E-06 | 5.1E-03 | 2.2E-04 | 2.4E-03 | 1.0E-06 | 4.0E-06 |
obscene | 2.0E-04 | 9.2E-03 | 1.7E-05 | 8.7E-01 | 1.7E-02 | 1.2E-01 | 1.7E-05 | 2.0E-04 |
threat | 7.0E-05 | 1.5E-03 | 1.2E-05 | 2.2E-01 | 2.6E-04 | 7.5E-03 | 1.3E-05 | 7.0E-05 |
insult | 8.4E-04 | 1.6E-02 | 7.6E-05 | 9.9E-01 | 1.5E-02 | 9.1E-02 | 7.9E-05 | 8.4E-04 |
Identity attack | 3.6E-04 | 3.8E-03 | 5.1E-05 | 6.9E-01 | 5.6E-04 | 9.3E-03 | 5.1E-05 | 3.6E-04 |
Sexual explicit | 6.2E-04 | 1.5E-02 | 7.0E-06 | 9.5E-01 | 2.7E-03 | 3.6E-02 | 7.0E-06 | 6.2E-04 |
