Research Article
ChemTok: A New Rule Based Tokenizer for Chemical Named Entity Recognition
Table 9
Class based performance (
-score in %) for SemEval DDI data set; DrugBank, Medline.
| Algorithm | Entity type | DrugBank | Medline | ChemSpot | tmVar | ChemTok | ChemSpot | tmVar | ChemTok |
| CRF | Group | 76.33 | 72.86 | 79.16 | 62.41 | 59.25 | 64.31 | Drug_n | 0.0 | 0.0 | 0.0 | 10.44 | 13.63 | 12.48 | Brand | 86.31 | 80.85 | 89.97 | 0.0 | 0.0 | 0.0 | Drug | 89.77 | 86.85 | 91.32 | 74.57 | 74.22 | 76.15 |
| SVM | Group | 83.82 | 83.58 | 85.12 | 46.28 | 44.06 | 49.13 | Drug_n | 0.0 | 0.0 | 0.0 | 10.93 | 11.02 | 11.67 | Brand | 92.15 | 94.11 | 93.45 | 0.0 | 0.0 | 0.0 | Drug | 91.66 | 89.32 | 93.52 | 68.04 | 67.06 | 71.71 |
|
|