1. Advertising
    y u no do it?

    Advertising (learn more)

    Advertise virtually anything here, with CPM banner ads, CPM email ads and CPC contextual links. You can target relevant areas of the site and show ads based on geographical location of the user if you wish.

    Starts at just $1 per CPM or $0.10 per CPC.

Semantics Test & First Results

Discussion in 'Google' started by T0PS3O, Jun 3, 2004.

  1. #1
    I separated these results from the thread here.

    I have set up a test a few days ago and have some results I wanted to share with you and discuss their significance.

    Three pages have been made with each the exact same number of words and characters. Each of them contain the key phrase 'tilting colema board' exactly 6 times in the paragraph text. On and off page SEO is the exact same. The only difference is the content to see if G understands what the pages are on about, and if so, does it affect ranking?

    Page X has pretty much random content, not related to the targeted key phrase at all.

    Page Y has lots of words related to the use of colema boards, as per my own imagination.

    Page Z is stuffed with words related to the keyphrase subject as suggested by G's very own Keyword Suggestion tool. If it does understand semantics and knows which words are related to which other words, then these suggestions are likely to come from the lexicon they have already built; being the suggestions and broad matches.

    Now the shocking first results (SERs):

    1: Page X (with B.S. text! :eek: )
    2: Page Z (with G's suggestions) :rolleyes:
    3: Page Y (with related words per my knowledge on the topic)

    Note: Page Y returns no rank in the Tracker.

    The page I made to get these three spidered does have Page X as the first link so maybe that might skew the results??? Input on this?

    See all three pages in G's results here: http://www.google.co.uk/search?q=si...ting+colema+board&hl=en&lr=&ie=UTF-8&filter=0

    Go read the text here: http://www.justbuyonline.co.uk/dev/test/semantics-test.html

    Could it be that the initial indexing only looks at exact KW matches? That maybe during the deep crawl and during the Dance the semantics factors come into the equation?

    Does G perhaps NOT even apply this at all? :confused:

    I'm not the kind of guy who draws quick conclusions but I have to say this is not near what I expected. It has only been in the SERPs for under 24hrs now so results might shuffle sooner or later. Would be interesting to see what happens if anything happens.

    I've read most of the papers suggested by compar and realize it's not an extensive test but what do you make out of this?
     
    T0PS3O, Jun 3, 2004 IP
  2. mxlabs

    mxlabs Peon

    Messages:
    327
    Likes Received:
    6
    Best Answers:
    0
    Trophy Points:
    0
    #2
    whats the page with google's keyword suggestions now? Y or Z?
     
    mxlabs, Jun 3, 2004 IP
  3. chachi

    chachi The other Jason

    Messages:
    1,600
    Likes Received:
    57
    Best Answers:
    0
    Trophy Points:
    0
    #3
    Wow, great effort Tops. The first thing that comes to mind is trying to isolate your first conclusion that at this early stage, G may be giving more relevance to the first link on the page. I would be happy to test that out on one of my sites for comparison.

    Maybe you should setup 2 other sections with similar pages, but order the links differently, then compare how they rank over time.

    It will be interesting to see if the pages with "related" keywords improve and rank better as time goes on.
     
    chachi, Jun 3, 2004 IP
  4. T0PS3O

    T0PS3O Feel Good PLC

    Messages:
    13,219
    Likes Received:
    777
    Best Answers:
    0
    Trophy Points:
    0
    #4
    Whoops, my bad! It is actually page Z. Let me change that.... Thanks for noticing.
     
    T0PS3O, Jun 3, 2004 IP
  5. compar

    compar Peon

    Messages:
    2,705
    Likes Received:
    169
    Best Answers:
    0
    Trophy Points:
    0
    #5
    I have all kinds of problems with tests like this. Lets say Google saw all these site as absolutley equal in relevance. How would they place them in the SERPs? So the order may mean absolutley nothing.

    The only conclusion I can draw from this test is that Google is not using any semantic tools in evaluating the content of the pages. The other problem with the test is that the use of semantic tools has always been discussed in terms of it being applied to pages on which IBL are placed. Never, to my way of thinking, to content of the page itself.

    The real test would be to put two links on each of these pages. One link would use "colema boards" as anchor text and the other some totally unrelated anchor text. Then try and evaluate the value passed by these link to the respective recipient page. I'm not sure how you would do that from pages with no PR like these.
     
    compar, Jun 3, 2004 IP
  6. mxlabs

    mxlabs Peon

    Messages:
    327
    Likes Received:
    6
    Best Answers:
    0
    Trophy Points:
    0
    #6
    regarding the topic I just think that google does not take semantic into account when ranking sites... of course it would be easy to realize, but we often overlook that google spiders a massive amount of websites. any "semantic AI" would eat CPUs for breakfast.

    and if google does filter out "nonsense grammar", what would exactly change? doorway generators and other spam programs would have a new feature by the end of the month: "use proper sentences".
     
    mxlabs, Jun 3, 2004 IP
  7. T0PS3O

    T0PS3O Feel Good PLC

    Messages:
    13,219
    Likes Received:
    777
    Best Answers:
    0
    Trophy Points:
    0
    #7
    Because of the possibility the rankings might still change I'd like to leave everything like it is for quite some time. Maybe this weekend I'll feel like repeating the same procedure on a different KeyPhrase and then indeed test the linking order. But still you can only tell when all 3 links go to the exact same page which on its turn might G make penalize some of them for similar content. That's a hard one to test really isn't it? How about making a 4th page 'Page W' similar to Page X with B.S. content and put it as a first link to and see if that one becomes #1 ? That might proof it.
     
    T0PS3O, Jun 3, 2004 IP
  8. T0PS3O

    T0PS3O Feel Good PLC

    Messages:
    13,219
    Likes Received:
    777
    Best Answers:
    0
    Trophy Points:
    0
    #8
    Very true. But wouldn't there always be even the slightest difference in relevance if G's algorithm is as complicated as we all keep believing it is?

    I find that a bit quick. There are tons of threads on the so-called 'sandbox effect' suggesting ranking does change when a site 'matures' so why could semantics not be part of this? I'm not saying it is but it is a bald statement to say it isn't.

    So you are saying semantics is only applied in the sence that the anchor text or perhaps the top KW's from the linking page are compared with the target page linked to and then semantics are used to compare relevancy?
     
    T0PS3O, Jun 3, 2004 IP
  9. disgust

    disgust Guest

    Messages:
    2,417
    Likes Received:
    133
    Best Answers:
    0
    Trophy Points:
    0
    #9
    interesting results

    one thing that you have to consider, though... even if google has these tools that are usually right, sometimes they won't be.. they have to weight that in. if the benefit from applying a tool is better than the negative effect it may have from the "wrong" intepretations of pages, then it's worthwhile.. but how would they measure something like that?
     
    disgust, Jun 3, 2004 IP
  10. compar

    compar Peon

    Messages:
    2,705
    Likes Received:
    169
    Best Answers:
    0
    Trophy Points:
    0
    #10
    But how do you know? Maybe they are all equal. You just haven't got sufficient data to draw any conclusion.

    To me the best sandbox theory is the one that suggests it is links that are put in the sandbox, not the pages. Remember it is links that drive the SERP placement -- not on-page stuff except for the most uniques content.

    Think about it this way. If Google immediately applied the full impact of IBLs, I could go out and buy 1,000 links and in 24 hours my page would leap forward in the SERPs. Then my competitors would go and buy 2,000 and with 24 everything would change. The SERPs would look like a yo-yo. So Google must be applying some dampening effect.

    The fact that all your pages showed up for your keywords would indicate that they were not put in any sandbox. And this makes sense because you don't have any real number of links to these pages.

    Ok, to start with I don't think semantics is being applied at all today. However, if it was, it would look at the anchor text of a link and compare it to the content of the page it appeared on and if it was thematically related the Algo would give the link more relevance.

    I don't think that it would look at the content of the target page. The evidence of the fact that Google works from anchor text, and not the target page's on-page content, is illustrated by all the successful Google bombings. These are still working and in almost every case the anchor text of the "bomb" has absolutely nothing to do with the content of the page it links to.
     
    compar, Jun 4, 2004 IP