There is so little context here that I cannot tell what this author was trying to measure, what their categories mean, or what conclusions they might draw from this experiment.
Generically I like the idea that you could identify attributes a text has and have a classifier attribute those attributes to documents and make some kind of report on it -- but this is so vague I just don't get it.