"In this benchmark, we extracted a list of the most common words in the book War and Peace, which contains over 500,000 words... Its accuracy was 100% in most cases, except for very high Ks where it ‘only’ achieved 99.9% accuracy"
Interesting how such probabilistic data structures can in most cases behave almost as good as deterministic.
