Multiple Choice Narrative Cloze (MCNC)
44.5AccuracySWCC
Evaluation Results
| Method | Links | |
|---|---|---|
| SWCCInference type=Unsupervised learning, Setting=Zero-shot transfer2022.03 | 44.5 | |
| BERT (InfoNCE)Inference type=Unsupervised learning, Loss Function=InfoNCE2022.03 | 39.23 | |
| Word2VecInference type=Unsupervised learning, Method=Skipgram summation of embeddings, Reporting source=Lee and Goldwasser, 20192022.03 | 37.39 | |
| BERT (Margin)Inference type=Unsupervised learning, Loss Function=Margin loss2022.03 | 36.5 | |
| PPMIInference type=Unsupervised learning, Method=Co-occurrence information and Positive PMI, Reporting source=Lee and Goldwasser, 20192022.03 | 30.52 | |
| BiGramInference type=Unsupervised learning, Method=Bi-gram conditional probabilities, Reporting source=Lee and Goldwasser, 20192022.03 | 29.67 | |
| RandomInference type=Unsupervised learning, Protocol=Random selection2022.03 | 20 |