conTest

Human-created dataset with high lexical diversity and varying semantic diversity for evaluating diversity metrics

Neighborhood — ranked by edge-count

paper