Table 3.

Accuracy of the automatic entity extraction process, based on author response

Data type	AVG Jaccard coefficient	AVG number of entities added per submission	AVG number of entities removed per submission
genes	0.85	0.38	1.23
species	0.75	0.28	0.36
alleles	0.88	0.26	0.43
strains	0.88	0.25	0.40
transgenes	0.95	0.04	0.09

The Jaccard coefficient measures the similarity between the set of automatically extracted entities and the set of entities submitted by authors through the feedback form.

AVG indicates average.