Table 3.
Benchmark datasets for WSD.
| Term | Manual (expert) | Manual (non-experts) | Semi-automatic | ||||
| False | True | False | True | False | True | ||
| GO | Development | 98 | 111 | 271 | 56 | 2296 | 715 |
| Spindle | 50 | 48 | 70 | 48 | 519 | 599 | |
| Nucleus | 99 | 100 | 25 | 61 | 131 | 1336 | |
| Transport | 102 | 91 | 102 | 56 | 1043 | 699 | |
| MeSH | Thrush | 17 | 83 | 45 | 7 | 35 | 1131 |
| Lead | 71 | 27 | 202 | 22 | 1564 | 735 | |
| Inhibition | 98 | 100 | 454 | 79 | 5247 | 553 | |
The above datasets contain manually collected PubMed articles by one expert (high quality/low quantity), manually curated articles by a group of non-experts (medium quality/medium quantity) and semi-automatically collected articles (low quality/high quantity). See Datasets section for details.