Figure 5From: Discriminative application of string similarity methods to chemical and non-chemical names for biomedical abbreviation clusteringThe distribution of the recalls (R), precisions (P) and F-measures (F) for the matchings of the chemical names (Chemical) or the non-chemical names (non-Chemical) obtained using SoftTFIDF with the threshold scale of 0.005 from 0.9 to 0.995. The x-axis corresponds to the threshold used to obtain matchings.Back to article page