A good word cloud algorithm could be built using pagerank type logic. Socially filter basic words frequency. Commodity words everybody uses should be small font. Words common in your milieu but uncommon elsewhere should be medium font. Words/catchphrases unique to you = big font.
Conversation
Replying to
Do you an example-pair of weak/default and useful clouds from the same corpus?
Replying to
I did this a few years back using a much less fancy technique (just divided the word freq in my writing corpus with the word freq from a standard corpus and sort high to low) omniorthogonal.blogspot.com/2015/01/whats- (didn't bother to wordcloud the results though).
1



