Jump to content

  • Log in with Facebook Log in with Twitter Log In with Google      Sign In   
  • Create Account

Subscribe to HRA Now!

 



Are you a Google Analytics enthusiast?

Share and download Custom Google Analytics Reports, dashboards and advanced segments--for FREE! 

 



 

 www.CustomReportSharing.com 

From the folks who brought you High Rankings!


Sponsored Content

 

 
 

Photo
- - - - -

Semantics


  • Please log in to reply
66 replies to this topic

#61 OldWelshGuy

OldWelshGuy

    Work is Fun

  • Moderator
  • 4,713 posts
  • Location:Neath, South Wales, UK

Posted 25 January 2004 - 04:09 PM

because there isn't enough corroborating evidence in Google's database to use semantic associations on that term

Which is why I do not appear on Googlisms, But George Bush does, maybe

#62 BrianR

BrianR

    Is it just me, or is it getting cooler in the evenings...?

  • Members
  • PipPipPipPipPipPipPip
  • 1,621 posts
  • Location:Chester, UK

Posted 25 January 2004 - 05:22 PM

That's why I say that I don't believe that there are two or more algos in play - there are just aspects of the algo that can't come into play for certain terms because there isn't enough corroborating evidence in Google's database to use semantic associations on that term.

Grumpus - perhaps you've said this before and I've missed it (a strong possibility!), but anyway, that makes a lot of sense to me - the idea that there is more than one algo in play doesn't feel right - it's not a clean solution - too messy! But then, what do I know about coding algos!

Yours, an interested observer...

#63 Grumpus

Grumpus

    HR 6

  • Active Members
  • PipPipPipPipPipPip
  • 786 posts

Posted 25 January 2004 - 06:15 PM

Shame on you, G. That's the kind of logic that resulted in people blithely sinking tons of cash into the Internet gold rush in the nineties because -- after all -- it must be a good investment if everyone else is doing it. I think we all know where that led.


Touche.

But it's not everyone jumping on a bandwagon with this. It's specific niche companies who see it working for them on some level.

We do know for a fact that some sort of semantical analysis is going on in Google because we can see that some of the things that look like stemming really aren't. So yes, it may not be being used at the levels we're describing it's potential to be, but it is in use at some level.

(I was going to show you a search result to demonstrate this, but it's a mess now with the update... it was a search where the "singular" of a word brought up medical information about the term whereas the plural brought up primarly legal and compensatory web sites. Therefore, it's not just a matter of stemming, but Google is actually assigning contextual meaning to terms. If it comes back, I'll demonstrate it, but it's not working with the update on, so....)


Maybe this whole process has to learn from hand-picked pages?


To some extent, maybe. When Google bought Applied Semantics, there was a dozen (or 2 dozen? for some reason the word dozen sticks in my head) employees whose job it was to watch and add their own associations and tweak the existing ones, etc. As far as I know, those folks are still employed.

the idea that there is more than one algo in play doesn't feel right



There are also several other parts unrelated to semantics that we can probably pretty safely assume are conditional. (i.e. If this is true, apply this factor, if not, apply another). This would give the illusion of it being completely separate algos, but it's not really.

G.

#64 SlickWilly

SlickWilly

    HR 1

  • Members
  • Pip
  • 3 posts

Posted 26 January 2004 - 02:32 PM

I know I am new to the forums but I wanted to add something to this discussion. Please forgive my formatting errors.

Has anybody looked at the site [http://vivisimo.com/]? The results come from a group of search engines (excluding google), but what impresses me is what it pulls for the categories on the sides. It seems like it is a step towards semantics even if it is not. They usually seem to be relevant and related to the search. Thisnews article has a mention of a possible partnership with google towards the end.

~SW

#65 BrianR

BrianR

    Is it just me, or is it getting cooler in the evenings...?

  • Members
  • PipPipPipPipPipPipPip
  • 1,621 posts
  • Location:Chester, UK

Posted 26 January 2004 - 06:25 PM

Welcome aboard, SlickWilly - hope you enjoy the fun!

There's been mention of vivisimo before on this forum - can't remember if it was in connection with semantics though - perhaps I'll do a topic search when I've finished posting...

BrianR

#66 nimlet

nimlet

    Where's that elephant got to?

  • Active Members
  • PipPip
  • 27 posts
  • Location:Bournemouth, UK

Posted 27 January 2004 - 07:04 AM

hey all,

I have read most of the articles etc. that have been posted - and that is a mighty stack of reading! Am now sufferring a small amount of brain ache and was hoping for a bit of clarification re all of this semantic stuff that I have been trying to get to grips with.

My site about holiday accommodation appears nowhere for the term "holiday accommodation", even though the term appears a fair bit in the text. When you search for "rent private holiday accommodation" we come up #22. The homepage appears in the serps but the phrase appears nowhere on the homepage/site.

Would that be a bit of semantics kicking in? as google can see that all of our content is related to renting private accommodation and doesn't need that exact phrase in the copy to bring our site up for that phrase?

I hope this doesn't sound really dumb, but I am trying to get it straight in my head - and it can take a while!

Cheers Nims

#67 Jill

Jill

    High Rankings Advisor

  • Admin
  • 32,311 posts

Posted 27 January 2004 - 09:10 AM

It's due to the competitive nature of the phrase, vs. the non-competitive nature of the other phrase.

Hundreds of thousands are competiting for holiday accomodation, and probably not too many for the longer phrase.

Jill




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users