Skip to main content

Table 6 Performance of Submitted Runs on the BioCreative III GN test corpus

From: Soft tagging of overlapping high confidence gene mention variants for cross-species full-text gene normalization

Annotation Run Precision Recall F-score TAP-5 TAP-10 TAP-20
  R1 0.4494 0.2316 0.3056 0.2137 0.2509 0.2509
test50.gold R2 0.4289 0.2352 0.3038 0.2086 0.2483 0.2483
  R3 0.4237 0.2364 0.3034 0.2099 0.2495 0.2495
  R1 0.8801 0.4136 0.5627 0.3820 0.3820 0.3820
test50.silver R2 0.8632 0.4316 0.5755 0.3855 0.3855 0.3855
  R3 0.8570 0.4360 0.5780 0.3890 0.3890 0.3890
  R1 0.8433 0.4327 0.5720 0.4540 0.4540 0.4540
test507.silver1 R2 0.8272 0.4377 0.5724 0.4536 0.4536 0.4536
  R3 0.8233 0.4427 0.5758 0.4577 0.4577 0.4577
  R1 0.9185 0.4743 0.6256 0.4873 0.4873 0.4873
test507.silver2 R2 0.9048 0.4818 0.6287 0.4871 0.4871 0.4871
  R3 0.9009 0.4875 0.6326 0.4916 0.4916 0.4916
  1. test50.gold: human annotation for the 50 articles
  2. test50.silver: pooled team submissions for the same 50 articles using the EM algorithm
  3. test507.silver1: human annotation for the 50 articles + pooled team results for the remaining 457 articles
  4. test507.silver2: pooled team submissions for all the 507 articles by the EM algorithm