tomaszjot

I know Google is obsessed with content. The question is: can G determine if text got sense or not?

Lets say article like:

Title: "Kitchen Equipment in Modern Flats"

Text: "I like my kitchen. Walls of the kitchen were yellow and green. Is oven still hot? Spoons in the drawer were full of dust." bla bla bla

Those sentences have no value to user that's obvious, but do they have any sense to Google? Just asking...

Please refrain from answering if you want to write something like "You should think about person reading it. Content is king and only unique and info reach content can bring happiness and money to you and your website users."

dchuk

each of those sentences makes sense on their own. Determining whether a piece of text is coherent and proves the point it sets out to prove is probably one of the more challenging AI problems of our time, and I would imagine Google wouldn't be able to do it yet

Aequitas

Google has been using LSI for years upon years, some of the most amazing efforts were placed into Google Wave which we know was a failed project but still technologically interesting.

I watched the engineers at Google talk about how they use linguistics and LSI to improve the quality of there search results but couldn't find it anymore.

Nonetheless the answer is simple, Google is exceptionally intelligent when it comes to semantic meaning of words and letters. When it comes to finding meaning in text the big G does exceptionally well better then almost anyone out there.

Lots of people don't take the time to realize that Google hires linguistic teachers just as they do engineers, both fields need to be combined in order to create a smarter better search engine.

Its a long complicated subject that'll take the next 20 years of research to fully explain so I'll just leave it at those few points.

__________________

tomaszjot

It is a great news for me Sir.

It is really interesting. I'm working on something, will see the final outcome, in a month or two I will know how good they are Thank you.

kyescontent

I used to work for the Google Rating Projects - where Google employed thousands of human beings to look over websites and decide if text made sense or if it was just keyword spam - or plain useless. Google laid off 10K people two years ago from the Rating project and removed the wanted ad from their site - but I was told by an employee that a version of the rating project still exists, through a different contractor.

dchuk

feel free to email me every single training manual and internal document you have and I'll archive them for you so you can free up some hard drive space

bauss

interesting first post

tomaszjot

Wonder if Adsense ads are the way of testing how Google understands the article?

If you just stuff article with keywords it doesn't have sense in Google's eyes. But how they will see article made from not-related sentences? All sentences have some sense alone. Let say whole text contains one or two main keywords within sentences. Then you mix it with sentences which are related with topic of article. So you have a mix which looks like this:

1-2 keyword rich phrases, 5-6 related phrases (but with different words) + some unrelated sentences

Wonder if that would work?

kyescontent

Believe me ... I looked at probably hundreds of sites that fell under this umbrella. We were instructed to mark them as "spam" for further review. I can't quite recall off the top of my head how our work related to the actual algorithms; but I do know that the algorithms were altered based partially on our ratings.

The rating guidelines were leaked onto the internet a few years ago by a disgruntled ex employee (I can hardly blame them; I've been unemployed since the mass lay-off I was part of!). Alas I looked for them just now and they seem to have been pulled down. I'm sure they're floating around somewhere though ... I don't believe I kept my copy.

K.

cougarclaws

did you review the site in a browser or look at the code?

Pranith

well google will determine text too.

If, the text in Article published by you is rich in content then your website will be on top. Apart from this meta tags.

I can show you few examples of content rich which on on google top page because of their content rich posts in their blogs or forums.

kyescontent

In a browser. The only times I looked at source were to look for hidden text.

kyescontent

You're quite right - Google puts a big emphasis on content keywords. As far as I know the search engine now ignores the keyword meta tag.

ii silver ii

Surely if they wanted a serious way to judge text content they could just analyse the amount of different types of words in one sentence to determine the weighting of adverbs, nouns, etc? Might take a long time could be affective?

__________________
'' ƛ '' _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
/∞\ | W | I | C | K | E | D | F | I | R | E |
\ ٮ / ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯

Pranith

Well having content with rich keywords doesn't take much time for google index. I done this practically for one of my clients website. make sure content is unique with rich keywords and 'header 1 tags' in content where ever necessary.

GoogleBitchSlap

I actually tested this with a keyword on one of my best ranking pages. I only added it to the meta tag keywords section, not anywhere else on the page. (it was a common british spelling for an english word, think extra e added)

Two weeks later when my page was reindexed by google, they dropped me on page two spot 11 for the keyword. ( note: I am pretty sure it is a low competition keyword since only brits use it that way)

I haven't created any backlinks with that anchor text. Its not in my h1 or content. Just in my meta tag.

To the OP:

If you are using this content / pages for a linkwheel or such just for SEO purposes, I would just go for it.

If you are using it on the site you want to make money on and rank well, just remember. In the long term, when you get to page 1 of the SERPs, will it pass manual inspection? If it wont, you are screwed. You will work hard just to get ranked then get slapped down and lose it all. This I know from experience :\

So it does affect it a bit in my humble researchings

Instrumentalist

This look familiar? http://www.huomah.com/quality-rater-guidelines-2007.pdf

lacrosse_20

Nice find. +rep.

__________________

kyescontent

I stand corrected! Thank you, that is incredibly valuable to know.

K.

kyescontent

Excellent, yes, that's at least part of it. I seem to remember it being longer, but maybe not. Thanks!

K.

kyescontent

... In fact, there were at least two sets of guidelines. There were General Guidelines, and Side-By-Side guidelines (one of the rating task types). There were also a number of tasks with more specific guidelines and experimental tasks which changed each time.

I'm pretty sure the current contractor is Lionbridge.

I'm still digging through my old files to see if I've got these documents anywhere, but regardless, if anyone has specific questions, I can probably answer them. The job was extremely repetitious, and I remember it well.

tomaszjot

Scrapped sentences still keep this balance right IMO. It could find out "cut" sentences.

3rd tier.

Wonder why the project died out so suddenly? (if) It could be because:

1. Google had no money (Google has a lot of money so no)
2. Proceed data was enough to create mega-spam-killing algorithm
3. Google's realized task is too big ever for them

Interesting, veeery interesting...

justo_tx

What do you think this is the matrix?

kyescontent

I'd love to know the answer to that. I do know that the job was always a temporary one year contract for everyone who ever worked it - and that people were routinely laid off and replaced. It was striking though that the project moved/was changed after so long (I'm not sure how many years, but it was quite a few). I know they used to employ 10K people, but I have no idea how many or few they contract now. The old contractor was WorkForceLogic - the new one is Lionbridge. WFL still handles the "ad quality rater" jobs, which I believe are a review of the paid/non-organic results.

K.

tomaszjot

So that's how the source code of your sites looks like? Oh my God, I'm so behind...

Maybe some of you guys used this research knowledge for a very-personal-projects?