homepage Welcome to WebmasterWorld Guest from 54.204.249.184
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Become a Pro Member
Visit PubCon.com
Home / Forums Index / Google / Google News Archive
Forum Library, Charter, Moderator: open

Google News Archive Forum

    
When did G start picking up plurals, possesives, etc
dealership = dealers, dealer, dealerships, etc
stcrim




msg:153963
 10:41 pm on Dec 25, 2003 (gmt 0)

Google use to be about exact word matches (not phrase but word matches)

My search for Ford Dealership {another Keyword} {another Keyword} returned every variation of every word including Ford's, Fords Dealers, Dealer, and on and on - each highlighted as though I had searched for it.

Is this new or have I just slept through something?

-s-

 

Marcia




msg:153964
 12:21 am on Dec 26, 2003 (gmt 0)

That's due to Google's use of stemming, which is fairly recent and was very quietly changed on their site recently, on the page where they previously had said that they didn't use stemming.

This lends itself to some interesting new opportunities, but IMHO can complicate things a little bit for those among us who are intent on optimizing pages for Inktomi.

rfgdxm1




msg:153965
 1:00 am on Dec 26, 2003 (gmt 0)

>This lends itself to some interesting new opportunities, but IMHO can complicate things a little bit for those among us who are intent on optimizing pages for Inktomi.

How does this complicate things?

GoogleGuy




msg:153966
 5:05 am on Dec 26, 2003 (gmt 0)

It's been several weeks. I think you can turn it off/do an exact search with + in front of the word, or by putting it in quotes.

yankee




msg:153967
 5:14 am on Dec 26, 2003 (gmt 0)

Google used to be like a rifle. You could find exactly what you were looking for. Now it's like a shotgun, due to stemming. If you don't know how to search with + (my guess is most people don't), it's much harder to find what you are looking for. In my opinion, google seriously degraded it's value by introducing stemming.

With old google, when you searched for kw1 kw2 you found exactly what you were looking for. Now, this same search will return results about kw2 that have very little to do with kw1. It's amazes me how google claims they have improved their search engine, when reality is they made it much worse!

Marcia




msg:153968
 5:32 am on Dec 26, 2003 (gmt 0)

Not totally, yankee. Not from a webmaster perspective because they can find our pages now for searches they couldn't have before, simply because we used one form of a word predominantly.

rfg, it's complicated because of the fact that with Ink there's a barebones simplicity of making pages that are very precisely targeted. With Google now you can be found by people looking for the alternate of a word - they can find what they're looking for either way when they search, whichever way they do it - in *some* cases.

If some pages are doing well with Ink, trying to vary them so they get found both ways with Google could mess up the balance just a little bit for the Ink listings. For some pages, anyway. It's a tomayto, tomahto thing. Some say it one way, some another but they're the same thing.

I've been getting Ink at Yahoo for several weeks and was looking at this very thing all afternoon, with two different forms of a certain item descriptor for the second word in two-word phrases and several different modifying words for the first word.

An interesting aspect is the phenomenon of the message we get that words are only found in "pages linking to this page," seeing what will happen with with different combinations of on-page and inbound link usage of the alternate words and phrases in varying proportions.

>>you can turn it off/do an exact search with + in front of the word, or by putting it in quotes.

There are some tremendous differences in numbers of pages returned, using the quotes or the +. For a two word phrase, using the + in front of each of the two words pulls in some very interesting difference.

I'm looking at one in particular. Normal "raw" search returns 4,590,000 pages - and the page in the #1 spot (not mine, btw) does not have that phrase on the page at all, nor in the site itself that I've seen because that "stuff" is not on that site at all, it's just a fancy doorway page.

Same search using the 2-word phrase in quotes returns 348,000 pages - with what's the #2 page in the first example coming up at #1 and not a sign of the page that was #1 in the first example.

Same search using +word1 +word2 brings back 4,620,000 pages returned with NO sign of the site that was #1 in the first, "raw" search above. Again, the site that was at #2 is #1 for this one, too.

In that search, it's word-2 that's the one that could be the variable, with either of two forms of the word. The site that's #2 and moved to #1 cannot rank for form #2 because the alternate form is not used, particularly not in anchor text, not even once.

For that stuff using the variable for the second word, the site (without the stuff) also comes up at #1 but whoooaaa!

Looking at the cache for that #1 site for both:

These terms only appear in links pointing to this page:

And for the #2 - indented result for the same site, same search

These terms only appear in links pointing to this page:

There are close to 4 million pages returned for the second variation - and in neither case is it because of what's on the page or the site. It's been all anchor text all along and still is. It was only the second form for a while, seems anchor text was added to pull in the traffic from the first search mentioned also.

On the other hand, that page is at the very end of top ten with Yahoo/Ink - by virtue of the description in the Yahoo Directory listing. There's not a snowball's chance in Miami mid-summer that site will hit #1 with Yahoo/Ink on its own. It would take some mighty high-handed finagling of the Directory to accomplish that as it stands now.

[edited by: Marcia at 6:05 am (utc) on Dec. 26, 2003]

IITian




msg:153969
 6:04 am on Dec 26, 2003 (gmt 0)

Does the concept of stemming apply to proper name variations too - like Bob and Robert, Joe and Joseph, Pat and Patricia and so on? Just wondered.

GoogleGuy




msg:153970
 5:26 pm on Dec 26, 2003 (gmt 0)

yankee, we've tested this a lot, and all our tests say it's an overall quality win for users. Very little in this world is perfect, so a few searches might get worse, but the large majority of searches get better. It took a while to find the right way to do this, so there were ways to do it that we tried that wouldn't have been a win or as much of a win; instead of deploying those, we went back to find a better way to do this. I agree that according to conventional informational retrieval wisdom, stemming would often be just a wash, but we've done some extra work.

IITian, name variants are not really stemming (e.g. coaching <-> coach), but more like synonyms. There may be some conservative amount of that going on right now, but not very much right now.

By the way, I'm traveling again over the next day or two, so I might not pop in as often for a while..

NFFC




msg:153971
 6:46 pm on Dec 26, 2003 (gmt 0)

>we've tested this a lot, and all our tests say it's an overall quality win for users

I'd be very tempted to review the testing methodology.

>according to conventional informational retrieval wisdom, stemming would often be just a wash, but we've done some extra work

A pig with lipstick on is still a pig ;)

On a more constructive note, if you are looking for unbiased feedback I'd be tempted to ask the guys and gals over at Google Answers their opinion. Spent some time over there today, seems they are having use some real complex search syntax to earn their dollars.

yankee




msg:153972
 10:17 pm on Dec 26, 2003 (gmt 0)

If all your tests say it's an overall quality win for users, then you need new tests. It's just a matter of time before google usage goes down. Google isn't any better than the other SE's any more. First time I've ever been able to say that.

zgb999




msg:153973
 2:39 pm on Dec 27, 2003 (gmt 0)

I am seeing advantages in stemming and synonyms BUT I doubt Google will be able so deal with all the problem of an individual searcher in parts of the world with different spellings than in other parts of the world. There are just too many regions / slangs or however you might call it to deal with.

There are so many words with different spellings in different parts of the world. If I am looking for a word using how it is used in my region of the world I preferr getting sites using the word exactly as I entered it in Google (as it used to be the case).

Bringing this technology to other languages than English will bring even much more pain than it is right now in some cases!

The advantage of Google is its ease of use. I don't think it is possible to train enough people to use + for their searches in order to find what they are looking for.

Miop




msg:153974
 2:52 pm on Dec 27, 2003 (gmt 0)

Results for me are bizarre - widgetry shop does not appear for 'shop' but does appear for 'shops' (right at the top of a list of shopping malls and link farms) even though it's only one shop! There isn't one other widgetry shop near me in the SERPS. Google appears to have linked it to my links page on which there is a list of 'widgetry shops'.

sudden




msg:153975
 4:00 pm on Dec 27, 2003 (gmt 0)

GG: Is stemming used for all languages - or just for queries in english?

Brett_Tabke




msg:153976
 5:54 am on Dec 29, 2003 (gmt 0)

[webmasterworld.com...]

Global Options:
 top home search open messages active posts  
 

Home / Forums Index / Google / Google News Archive
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved