[Snowball-discuss] restoring a stemmed word

Arjen van der Meijden arjen@glas.its.tudelft.nl
Fri Jul 4 14:09:01 2003


> Alexandra Elizabeth Duncan wrote:
> 
> I haven't seen any mention of ways to restore a stemmed word 
> and I was 
> wondering if there are similar rules for doing so, and/or can one 
> reverse the algorithm to restore a word back to its full form?
Afaik, that is impossible. You could consider stemming to be a lossy
compression... So you really loose knowledge of the original words which
can't be deducted from the stemmed form.

> Any advice or pointers to reading material on this would be greatly 
> appreciated.
There are probably very efficient compression rules for wordtables, also
were a prefix-based (of all words with the same prefix, only store the
difference/additional characters and the prefix once) could work out.

But I'm absolutely no expert in this field.

Regards,

Arjen