Validation: Skewness of entity/topic distributions.

[ampudia19]

Note that a first set of histograms doing just this (and using highest-ranking entities to label topics) exists in issue #12

[emily-bicks]

I actually think entropy is a better measure than skewness, because we're not comparing to a normal distribution but rather a uniform distribution