Google's similarity algorithm - GoFuckYourself.com

debbieN · 08-12-2005, 11:49 AM

Does anyone have any information on this.

I have multiple category pages under one domain. What does Google look for in similar attributes on your pages? Similar to the main domain page or all the sub pages together?

If it just body text, then it is easy to keep from getting a lowered PR.

I think it is much more than that. Page size (71K for all lower pages), scripts, meta language, word patterns in text, who knows this.

GoogleGuy - is she/he for real?

Sosa · 08-12-2005, 11:55 AM

do you change titles, meta tags, text on the page? Or is all the content the same except for the keywords?

SmokeyTheBear · 08-12-2005, 11:58 AM

not quite sure what you mean with the "similar" thing but..

If you mean when it finds 86 of 300 and the rest are the same it just means it already has a page in its cache that is quite similar from your domain.

if you want to keep them seperate make them subdomains

debbieN · 08-12-2005, 12:07 PM

Sosa,

the content on each page is totally different.

debbieN · 08-12-2005, 12:08 PM

Smokeythebear

thanks,

you just told me what I need to know.

wedouglas · 08-12-2005, 01:09 PM

It has to do with content. Otherwise dynamic pages would be considered too similiar

dcortez · 08-12-2005, 01:33 PM

I've been observing the similarity filters and dup content thresholds on G and other SE's over several generations of algos.

G's algos for flagging pages as being similar are quite sophisticated. For example, pages made up of small snippets from other pages, however random, at some point start being seen as similar.

This applies to links pages which use the same titles and descriptions (as everyone else). Those titles and descriptions get 'burned' and they are not considered unique any more (when they are all over the web).

The most assured way to minimize tripping dup/similar filters is to write each page's text as a new piece of work. If you have multiple pages which link to a common page and your links include descriptions, use original, new descriptions for each link on every page. Even so, when you start cranking out hundreds and thousands of pages, statistically, you will start repeating and while you may not see it from within your site, an algo can trip a filter across large numbers of pages.

In summary, a web page is considered (by SEs) as:

a) content page - when most of the text content on the page is original (not repeated anywhere else on your site or others) and with minimal off-site links.

b) hub page - when there is a high ratio of off-site links to actual original text content. Link lists are examples of hub pages.

c) both - when there is a mix of off-site links and original content.

From my experience, it seems to be best to create either content or hub pages. They are more effective for what they are.

In any case, when linking to internal and offsite pages, it is important that you link to pages which relate to the theme of the page you are linking from.

-Dino

Linkster · 08-12-2005, 02:38 PM

Debbie - in my experiences - and I have many "category" pages that rank #1 for their niche or at least in the top 10 for highly competitive phrases - the difference should be around 15-25%, not counting the outgoing links - if you have around 200 or so links out going and a good amount coming in the percentage can be a little lower.

As far as Googleguy - yes he is real - he is a self described engineer that works for Google and most think he is Matt Cutts - he has a blog here: http://www.mattcutts.com/blog/ where he describes his responses to different types of schemes and SEO tricks

08-12-2005, 11:58 AM	#3
SmokeyTheBear ►SouthOfHeaven Join Date: Jun 2004 Location: PlanetEarth MyBoardRank: GerbilMaster My-Penis-Size: extralarge MyWeapon: Computer Posts: 28,609	not quite sure what you mean with the "similar" thing but.. If you mean when it finds 86 of 300 and the rest are the same it just means it already has a page in its cache that is quite similar from your domain. if you want to keep them seperate make them subdomains __________________ hatisblack at yahoo.com

08-12-2005, 12:07 PM	#4
debbieN Confirmed User Join Date: Jan 2003 Location: FL, NC(I burned this down), MI Posts: 1,193	Sosa, the content on each page is totally different. __________________ don't rent the top room at the Luxor ______________________ Debbie Nomad's Evil Hobby

08-12-2005, 12:08 PM	#5
debbieN Confirmed User Join Date: Jan 2003 Location: FL, NC(I burned this down), MI Posts: 1,193	Smokeythebear thanks, you just told me what I need to know. __________________ don't rent the top room at the Luxor ______________________ Debbie Nomad's Evil Hobby

08-12-2005, 02:38 PM	#8
Linkster Confirmed User Join Date: Feb 2003 Location: DeltaHell Posts: 3,216	Debbie - in my experiences - and I have many "category" pages that rank #1 for their niche or at least in the top 10 for highly competitive phrases - the difference should be around 15-25%, not counting the outgoing links - if you have around 200 or so links out going and a good amount coming in the percentage can be a little lower. As far as Googleguy - yes he is real - he is a self described engineer that works for Google and most think he is Matt Cutts - he has a blog here: http://www.mattcutts.com/blog/ where he describes his responses to different types of schemes and SEO tricks __________________ Porn Links Crak for your bank account - Get Addicted FREE LIVE CAM SHOWS

08-12-2005, 11:49 AM	#1
debbieN Confirmed User Join Date: Jan 2003 Location: FL, NC(I burned this down), MI Posts: 1,193	Google's similarity algorithm Does anyone have any information on this. I have multiple category pages under one domain. What does Google look for in similar attributes on your pages? Similar to the main domain page or all the sub pages together? If it just body text, then it is easy to keep from getting a lowered PR. I think it is much more than that. Page size (71K for all lower pages), scripts, meta language, word patterns in text, who knows this. GoogleGuy - is she/he for real? __________________ don't rent the top room at the Luxor ______________________ Debbie Nomad's Evil Hobby

08-12-2005, 11:55 AM	#2
Sosa In Tushy Land Join Date: Oct 2002 Location: Nebraska Posts: 40,149	do you change titles, meta tags, text on the page? Or is all the content the same except for the keywords?

08-12-2005, 01:09 PM	#6
wedouglas So Fucking Banned Join Date: Aug 2004 Location: Dearborn, MI Posts: 5,921	It has to do with content. Otherwise dynamic pages would be considered too similiar

08-12-2005, 01:33 PM	#7
dcortez DINO CORTEZ™ Industry Role: Join Date: Jun 2003 Location: Vancouver Island Posts: 2,145	I've been observing the similarity filters and dup content thresholds on G and other SE's over several generations of algos. G's algos for flagging pages as being similar are quite sophisticated. For example, pages made up of small snippets from other pages, however random, at some point start being seen as similar. This applies to links pages which use the same titles and descriptions (as everyone else). Those titles and descriptions get 'burned' and they are not considered unique any more (when they are all over the web). The most assured way to minimize tripping dup/similar filters is to write each page's text as a new piece of work. If you have multiple pages which link to a common page and your links include descriptions, use original, new descriptions for each link on every page. Even so, when you start cranking out hundreds and thousands of pages, statistically, you will start repeating and while you may not see it from within your site, an algo can trip a filter across large numbers of pages. In summary, a web page is considered (by SEs) as: a) content page - when most of the text content on the page is original (not repeated anywhere else on your site or others) and with minimal off-site links. b) hub page - when there is a high ratio of off-site links to actual original text content. Link lists are examples of hub pages. c) both - when there is a mix of off-site links and original content. From my experience, it seems to be best to create either content or hub pages. They are more effective for what they are. In any case, when linking to internal and offsite pages, it is important that you link to pages which relate to the theme of the page you are linking from. -Dino