kawu / adict Goto Github PK

Check if there is a queue implementation which will be faster than Data.Set for the shortest-path algorithm.

Allow for cost customization:

• Cost depends on position
• Insertion and deletion depend on position and character

Forward and backward trie method for Levenshtein search.

Implement extended edit distance searching method. Speed of the method is of no concern to us right now.

Optimize Levenshtein search:

• Memory usage (use mutable dynamic table)
• Compute only diagonal values

Implement a strict insert function. Then, Trie construction will be a matter of simple fold over the input list.

Some functions work on a Trie data structure. It seems that it is always a better idea to use DAWG rather than Trie, so we can rewrite such functions to work on a DAWG.

• Generalize alphabet element type from Char to any, arbitrary type
• (ListLike w c => w) instead of String or Text

Test different search functions with QuickCheck.