One of the most efficient ways (that I know of) is to create a trie [0] from your corpus, then incorporate the levenshtein algo into the tree traversal algo to

One of the most efficient ways (that I know of) is to create a trie [0] from your corpus, then incorporate the levenshtein algo into the tree traversal algo to give you a levenshtein automata over a trie. Since a trie is a binary tree, if it's properly balanced, this timing of this operation is O(log(n)). A less efficient way is to traverse the trie and calculate the Levenshtein distance for each word you encounter. An even less efficient solution might be to sort all strings in your corpus by length and calculate distance for all of them. [0] https://en.wikipedia.org/wiki/Trie https://en.wikipedia.org/wiki/Trie