Table 21.
Exact match results for the concept normalization experiments on the core evaluation annotation set of 30 held-out documents (class ID level)
| Ontology | Token-ids (%) | Type-ids (%) | Shuffled-ids (%) | Random-ids (%) | Alphabetical-ids (%) |
|---|---|---|---|---|---|
| ChEBI | 82* | 65 | 0 | 0 | 78 |
| CL | 72* | 72* | 69 | 70 | 56 |
| GO_BP | 82* | 79 | 64 | 27 | 52 |
| GO_CC | 81 | 80 | 81 | 84* | 76 |
| GO_MF | 98* | 98* | 98* | 98* | 51 |
| MOP | 95 | 97* | 97* | 92 | 80 |
| NCBITaxon | 87* | 0 | 0 | 0 | 0 |
| PR | 10* | 3 | 0 | 0 | 8 |
| SO | 97* | 96 | 97* | 97* | 96 |
| UBERON | 78* | 74 | 0 | 0 | 74 |
We report the exact match percentage at the class ID level. The highest percentage is bolded and with an asterisk*