K-means properties on six clustering benchmark datasets

I propose the index h, defined as the number of papers with citation number > or =h, as a useful index to characterize the scientific output of a researcher.