Re: new version of spell checker

[David Spencer <da...@tropo.com>]

Nicolas Maisonneuve wrote:

> UPDATE
> - sort fixed (the sort was inversed!) 
> - set gram dynamicaly (depending of the length of the word) 
> - use the FuzzyQuery score: ((edit distance)/(length of word))
> - new Dictionary interface + LuceneDictionary  and PlaintextDictionary implementation
> - replace addWords method by indexDictionary(Dictionnary dic)
> - add  a new public method: boolean exist(word) 
> - add a build.xml
> 
> see the wiki page http://wiki.apache.org/jakarta-lucene/SpellChecker
> 
> 1 - Could we put the spellchecker to the sandbox.. it'll be easier to maintain than use Bugzilla/wiki process ?

Hi - I'm the orig author of this and a month ago got write access to the 
sandbox - then my wife had our 2nd child and I got distracted w/ child 
rearing...and I always intended to check in "v1.0" of the spell checker 
I contributed - but since we're up to v1.1 the least I could do is check 
it in now so voilla, I have checked it in.

I tried to make minimal changes in this checkin so it reflects the 
sources in the zip file on the wiki (URL above).

I have verified that it compiles but that's all - I have not tested it 
at all, and I have not even run the unit test as I couldn't get it to 
compile w/ junit, but I didn't try that hard.

TBD is that I think we should use the "include ../common.xml" idiom that 
I think other top level build files use.

Could someone verify that when checked out this builds at least?
I think you'll have to set the location of the lucene jar a la "ant 
-Dlucene.lib=...."

You sould see this under jakarta-lucene-sandbox/contributions/spellchecker


I guess another TBD is to review the package the code is in - does 
org.apache.lucene.search.spell make sense? I'd put it up a level ( thus 
get rid of the 'search' component).

Also, rambling on, another TBD for me at least is to compare it to my 
orig humble contribution as the code seems to have been reworked.

thx for the enhancements and getting it into the proper dir tree,

- Dave


> 
> 2 - Jonathan Hager: Could you test this version with our dictionary and said me the results ?
> 
> 3 - I search a french dictonary , someone has a URL where i could download it ?
> 
> thanks to Jonathan Hager, and Aad Nales for your suggestions / observations ;-)
> 
> Nicolas Maisonneuve
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: lucene-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: lucene-dev-help@jakarta.apache.org