Skip to main content

Table 5 IMT macro-averaged participant results

From: The Protein-Protein Interaction tasks of BioCreative III: classification/ranking of articles and linking bio-ontology concepts to full text

Team Run/Srvr Docs Precision Recall F-Score AUC iP/R
Base _Top4 RUN _1 222 28.60 47.75 33.78 0.32771
Base _Regex RUN _1 153 33.88 22.31 24.73 -
T65 RUN_1 222 9.35 83.21 16.32 0.47884
T65 RUN_2 222 2.45 100.00 4.75 0.44034
T65 RUN_3 222 9.99 79.38 17.16 0.47650
T65 RUN_4 222 33.48 42.88 35.40 0.30927
T65 RUN_5 222 2.44 100.00 4.74 0.50111
T69 RUN_1 214 54.87 57.91 52.39 0.52112
T69 RUN_2 211 57.01 57.35 53.42 0.51844
T69 RUN_3 203 60.24 56.41 54.45 0.51470
T69 RUN_4 199 62.46 55.17 55.06 0.51013
T69 RUN_5 190 64.24 52.44 54.35 0.49390
T70 RUN_1 143 51.78 35.01 37.84 0.31402
T70 RUN_2 72 71.76 36.81 45.61 0.36215
T70 RUN_3 30 80.00 41.50 51.51 0.41500
T70 RUN_4 205 31.65 38.72 31.75 0.32295
T70 RUN_5 159 36.36 21.26 24.75 0.18976
T81 RUN_1 222 4.44 63.91 8.19 0.22022
T81 RUN_2 221 9.39 41.92 14.12 0.19766
T81 RUN_3 222 13.51 28.35 17.41 0.17010
T81 RUN_4 222 13.21 29.57 17.34 0.20388
T81 RUN_5 209 21.93 24.64 21.34 0.18733
T88 RUN_1 219 29.10 45.04 33.60 0.38590
T88 RUN_2 220 28.67 45.53 33.35 0.38373
T89 RUN_1 200 54.78 53.37 50.91 0.46061
T89 RUN_2 200 54.95 53.23 50.76 0.46423
T89 RUN_3 201 54.05 53.25 50.23 0.45330
T89 RUN_4 199 54.48 54.18 51.25 0.47211
T89 RUN_5 201 55.30 56.12 52.38 0.47807
T89 SRVR_4 200 55.33 55.61 52.11 0.47636
T89 SRVR_5 199 54.09 54.00 50.96 0.47650
T89 SRVR_6 201 55.14 56.12 52.35 0.48047
T89 SRVR_7 203 50.46 55.66 50.06 0.47392
T89 SRVR_8 199 54.04 54.05 50.84 0.47534
T90 RUN_1 200 56.11 51.59 50.72 0.44687
T90 RUN_2 203 56.37 53.19 51.20 0.47159
T90 RUN_3 217 55.29 59.90 54.62 0.52974
T90 RUN_4 177 63.98 46.89 51.36 0.44118
T90 RUN_5 164 66.26 46.78 52.02 0.44458
T100 RUN_1 213 47.26 54.97 47.06 0.43312
T100 RUN_2 222 41.19 54.61 44.18 0.43238
T100 RUN_3 222 35.29 45.53 37.50 0.32459
T100 RUN_4 222 35.29 45.53 37.50 0.32459
T100 RUN_5 125 56.40 30.65 37.01 0.29387
  1. Macro-averaged results when evaluating only documents for which the system reported results (i.e., measuring the average per-document performance only on the documents each run produced annotations for). The highest score for each evaluation column is show in bold typeface, the lowest in italics. Run/Srvr: RUN = offline run, SRVR = online server run via BCMS; Docs: number of documents annotated; AUC iP/R: Area under the interpolated precision/recall curve. Base Top4: baseline system that assigns the four most frequent classes ordered by their frequencies in the training/development set. Base Regex: simple matching strategy based on regular expressions.