Defective Dialog Detection on OOD Shopping n = 105 (test)
48PrecisionDQM
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DQMmodel_type=supervised model2023.06 | 48 | 80 | 60 | |
| Last-turn TLDaggregation_strategy=last-turn2023.06 | 47 | 23 | 31 | |
| Rising linear weightsaggregation_strategy=linear weighting2023.06 | 41 | 67 | 51 | |
| Mean TLDaggregation_strategy=mean2023.06 | 39 | 77 | 52 | |
| Union of mean & last-turn TLDaggregation_strategy=union2023.06 | 38 | 77 | 51 |