6 ms·
From the article: If you’re using a small corpus and long Markov chains, you’ll end up with lots of actual strings from the corpus, and no fake ones. If this h
by a_m0d 15y ago
From the article:
If you’re using a small corpus and long Markov chains, you’ll end up with lots of actual strings from the corpus, and no fake ones. If this happens, experiment with the second parameter to the constructor for the class “MarkovGenerator.”
For the authors you are using, the corpus is too small.
- vedant 15y agoSince Paul Erdos is one of the most prolific mathematicians ever, the problem is more likely that the default chain length (4) is too long for even a relatively large corpus. A chain length of 2 does much better.