Question

I’m looking to write an efficient n-order Markov chain method to generate random text strings given a set of example text. I currently have a Java implementation that uses several layers of Maps, but it’s clunky. A suffix array would be perfect for my needs, but I’m not clear if that can be efficiently implemented in Java.

In C I might do something like:

char exampleText[MAX];
char *suffixArray[MAX];
...
while(n<MAX && suffixArray[n++] = &exampleText[n]);
sort(suffixArray);

This gets gnarly in Java since I’d have to take substrings of exampleText, or turn suffixArray into an array of indices, or something else.

Any suggestions for a good approach to this in Java?

Answer 1

String will [typically] do that for you. (Typical implementations share backing arrays when created with substring, although that is subject to change at any time.)