Skip to main content

Table 2 Performance of NER.

From: Automatic recognition of topic-classified relations between prostate cancer and genes using MEDLINE abstracts

Target Entities Features Precision (%) Relative recall (%)
  1 2 3 4 5 6 7   
GENE         84.4 100.0
         93.5 95.4
         95.0 97.6
         93.1 93.3
         84.4 100.0
         84.4 100.0
         84.4 100.0
        84.4 100.0
       95.8 97.0
PROSTATE CANCER         99.2 100.0
         99.3 99.8
         99.3 100.0
         99.3 100.0
         99.2 100.0
         99.2 100.0
         99.2 100.0
        99.2 100.0
       99.3 100.0
  1. Note: 1) Bag of words (all words in co-occurrence), 2) candidate gene and prostate cancer names, 3) unigram words, 4) presence of capital letters in candidate term, 5) presence of numerical digits in candidate term, 6) presence of Greek letters in candidate term, 7) presence of affixes of candidate term.
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us