Attribute Value Extraction on AE-pub cleaned (test)
47.74Macro PrecisionBERT-QA +vals +drop +mixing
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| BERT-QA +vals +drop +mixingModel variant/components=Knowledge-driven query expansion + Knowledge dropout + Knowledge token mixing, Pre-trained backbone=BERT-base, Number of trials=52022.06 | 47.74 | 44.82 | 46.23 | 87.84 | 82.61 | 85.14 | |
| BERT-QA +vals +mixingModel variant/components=Knowledge-driven query expansion + Knowledge token mixing, Pre-trained backbone=BERT-base, Number of trials=52022.06 | 46.67 | 43.32 | 44.93 | 88.3 | 82.46 | 85.28 | |
| BERT-QAPre-trained backbone=BERT-base, Number of trials=52022.06 | 42.77 | 40.85 | 41.79 | 87.14 | 82.16 | 84.58 | |
| AVEQANumber of trials=52022.06 | 41.93 | 39.65 | 40.76 | 86.95 | 81.99 | 84.4 | |
| BERT-QA +vals +dropModel variant/components=Knowledge-driven query expansion + Knowledge dropout, Pre-trained backbone=BERT-base, Number of trials=52022.06 | 41.61 | 38.22 | 39.84 | 88.46 | 82.02 | 85.12 | |
| BERT-QA +valsModel variant/components=Knowledge-driven query expansion, Pre-trained backbone=BERT-base, Number of trials=52022.06 | 39.48 | 35.6 | 37.44 | 88.82 | 81.77 | 85.15 | |
| DictionaryNumber of trials=52022.06 | 33.2 | 30.37 | 31.72 | 73.39 | 73.77 | 73.58 | |
| SUOpenTagNumber of trials=52022.06 | 30.92 | 28.04 | 29.41 | 86.53 | 79.11 | 82.65 |