Research Article
[Retracted] MKA: A Scalable Medical Knowledge-Assisted Mechanism for Generative Models on Medical Conversation Tasks
Table 3
Comparison of the models on MedDialog-CN, where the best results are in bold.
| Model | Dialog-GPT | Transformer | BERT-GPT | MKA-Transformer | MKA-BERT-GPT |
| Perplexity | 9.71 | 9.52 | 8.23 | 8.81 | 8.04 | BLEU-2 | 5.21% | 4.92% | 4.88% | 5.02% | 5.71% | BLEU-4 | 1.83% | 0.90% | 0.97% | 0.99% | 1.35% | NIST-2 | 0.36 | 0.42 | 0.40 | 0.43 | 0.44 | NIST-4 | 0.32 | 0.40 | 0.39 | 0.40 | 0.43 | METEOR | 12.32% | 13.11% | 12.83% | 13.4% | 13.94% | Entropy-4 | 13.73 | 13.51 | 13.8 | 13.72 | 14.1 | Dist-1 | 0.02% | 0.03% | 0.03% | 0.03% | 0.04% | Dist-2 | 2.01% | 2.02% | 2.14% | 2.11% | 2.22% |
|
|