Fig. 5From: The performance of BERT as data representation of text clusteringBoxplot of the first 20 AG News data representation from BERT + Mean (a) before min–max normalization (b) after min–max normalizationBack to article page