Skip to main content

Table 3 Number of false positives and true positives at different levels of consensus from best micro-averaged runs of the 20 teams

From: Overview of BioCreative II gene normalization

Votes Count FP Count TP Precision Recall F-measure
20 1 86 0.989 0.110 0.197
19 3 204 0.986 0.260 0.411
18 7 288 0.976 0.367 0.533
17 8 359 0.978 0.457 0.623
16 11 421 0.975 0.536 0.692
15 13 470 0.973 0.599 0.741
14 15 513 0.972 0.654 0.781
13 19 555 0.967 0.707 0.817
12 30 572 0.950 0.729 0.825
11 42 599 0.934 0.763 0.840
10 51 623 0.924 0.794 0.854
9 77 644 0.893 0.820 0.855
8 103 667 0.866 0.850 0.858
7 130 685 0.840 0.873 0.856
6 160 704 0.815 0.897 0.854
5 221 714 0.764 0.910 0.830
4 304 721 0.703 0.918 0.797
3 435 743 0.631 0.946 0.757
2 713 751 0.513 0.957 0.668
1 2522 763 0.232 0.972 0.375
Total   785    
  1. The table shows cumulative number of false positives and true positives (columns 2 and 3) obtained for a given level of consensus (column 1) from the top micro-averaged run of each team. Recall, precision, and F-measure were calculated using the consensus level as the minimum number of votes needed to include an identifier as an 'answer'. The total under True Positive Count indicates that there were 22 true positives that no system identified; see additional data file 3 for a listing of these.