Phil Tomson wrote:
> In article <d0fd5d6105092210597d704f9e / mail.gmail.com>,
> Josh Charles  <josh.charles / gmail.com> wrote:
> 
>>On 9/22/05, Phil Tomson <ptkwt / aracnet.com> wrote:
>>
>>>What kind of ai are you trying to integegrate with your web app?  Or mayb=
>>
>>e I
>>
>>>should ask: what do you want the ai to do for you?
>>
>>Well, I have a few ideas that would effect usability and create a
>>better end-user experience, but I'm still in the process of deciding
>>whether or not the costs of implementing the system would be worth the
>>benefits. Also, many of the features could also be implemented using a
>>rules based architecture, and I know if that would be considered AI or
>>not.
>>
>>It is similar to googles "did you mean aardvark?" when you search for
>>"ardvark."  is that considered AI?
> 
> 
> Good question. It may not be.  Could be simply that they check each of the 
> words against a dictionary and it it's not there they check for similar words 
> (using Levenstein distance (sp?), for example).  May not be all that much AI 
> going on there.
> 
> Phil
> 
> 
Hello all.
I come in not as a Ruby programmer,I just follow your postings from a 
Python programmer POV :)

I'm working on the fuzzy intelligence for that sound-like service since 
a month.

Given that a huge db of edit-distances between words is probably 
practical on some machines (I suppose a table 500000x500000/2 of floats 
would give some results), what I'm building is a clusterer for words to 
subdivide the words population in small sets each linked to some 
neighbours ,having only small qgram-distance words inside.
For what it's worth clustering involve AI as it's a simple form of training.

Ciau all, have fun


	

	
		
___________________________________ 
Yahoo! Mail: gratis 1GB per i messaggi e allegati da 10MB 
http://mail.yahoo.it