Re: [Algorithms] Dictionary compression

[Suncica H. <su...@ar...>]

When I wrote a spell checker many years ago I extracted common prefixes and 
suffixes from words and stored those in look-up tables. I used up to 3 
letters for prefix and same for suffix. With some manual tuning I have 
extracted stems very well and made words much shorter to search for using 
the algorithm similar to Trie. The number of different words was also 
reduced (example: create/pre-create, tree/sub-tree, test-ing/test-ed/test-s 
in English), so it took less memory. The spell checker was very fast.

Hope this helps,
Sunny