Fig. 6

Individual ratings distribution. This chart shows the distribution of annotators’ ratings per similarity grade during the evaluation of the corpus (not the training period). The x-axis shows the five similarity scale values, and the percentage of evaluations within each grade are represented on the y-axis. More than 40% of the evaluations were rated as “no similarities” (score of 0); nevertheless, 50% of evaluations were in the similarity value range between 1 and 3