Skip to main content

Table 3 Performance (in %) of individual systems on the training material, before and after stop-word removal.

From: Recognition of chemical entities: combining dictionary-based and grammar-based approaches

  Baseline Baseline + stop-word removal
  Precision Recall F-score Precision Recall F-score
Dictionary-based       
   ChEBI 28.3 40.6 33.4 77.7 39.7 52.6
   ChEMBL 87.9 18.7 30.8 88.8 18.7 30.9
   ChemSpider 65.4 39.0 48.9 80.4 38.4 51.9
   DrugBank 63.0 17.2 27.0 78.1 17.1 28.1
   HMDB 53.2 34.5 41.8 81.3 33.9 47.9
   NPC 46.8 26.7 34.0 59.7 26.4 36.6
   TTD 43.9 14.7 22.1 82.9 14.4 24.6
   PubChem 17.4 59.0 26.9 61.1 57.9 59.5
   Jochem 64.2 52.5 57.8 67.1 52.5 58.9
   UMLS 37.7 51.1 43.4 45.4 50.8 47.9
   ChEBI Family 10.4 16.6 12.8 29.4 16.3 21.0
Grammar-based       
   Oscar 25.1 63.2 35.9 28.4 62.4 39.0
   LeadMine 64.9 47.4 54.8 74.6 47.1 57.7
   ChemAxon 80.9 41.8 55.1 82.5 41.7 55.4
  1. The highest score in each column is bolded.