Re: [edict-jmdict] aruyouni

[René Malenfant <rene_malenfant@***********>]

I don't know what ちゃせん is, but I'm rather sure that Google has  
some kind of corpus that it uses to parse input strings into  
constituent words.

Of course I could be dead wrong about this.  This is what Jim and  
Paul specialize in.


Rene


On 18-Feb-08, at 9:44 PM, Jim Rose wrote:

> What your suggesting then is that an unquoted string is run through  
> an analysis such as ChaSen before being run through a database?   
> Otherwise how would Google know to parse that string into three words.
>
>
>
>
>
> On Feb 18, 2008, at 8:29 AM, René Malenfant wrote:
>
> > Well, that comment was directed at Paul, but I just tried what you
> > suggested and I got the same 140,000,000 result as you.
> >
> > The [G] link does not appear to be putting the search string in
> > quotation marks. i.e., it looks for あるように, not "あるよ
> > うに", so it drastically overestimates the number of hits. (AFAICT,
> > it searches for any page that has ある、よう and に, but not
> > necessarily as "あるように". With three such common words, it's
> > basically returning every Japanese page on the web.)
> >
> > If there's no technical difficulty preventing it, perhaps the [G]
> > links should be changed to use quotation marks in their search  
> > strings?
> >
> > Rene