Skip to main content

Table 3 Performance of relation recognition.

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

  Features Precision (%) Relative recall (%)
  Order of entities Bag of words Candidate entities Unigram Bigram   
Any relation 0.907 0.958
   0.908 0.957
    0.896 0.975
    0.902 0.957
     0.903 0.966
    0.909 0.961
Study description 0.669 0.567
   0.650 0.568
    0.597 0.295
    0.662 0.551
     0.659 0.552
    0.651 0.550
Genetic variation 0.793 0.691
   0.781 0.665
    0.737 0.464
    0.791 0.669
     0.796 0.687
    0.794 0.680
Gene expression 0.734 0.614
   0.720 0.620
    0.762 0.466
    0.733 0.612
     0.743 0.613
    0.735 0.612
Epigenetics 0.854 0.660
   0.833 0.660
    0.905 0.358
    0.857 0.679
     0.854 0.660
    0.854 0.660
Pharmacology 0.657 0.431
   0.626 0.442
    0.642 0.264
    0.643 0.419
     0.637 0.419
    0.647 0.433
Clinical marker 0.774 0.723
   0.767 0.728
    0.772 0.730
    0.768 0.678
     0.771 0.727
    0.772 0.719
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us