Clues To Google's Algorithm?

Garrett · #1 (**permalink**) 01-26-2004, 11:44 AM

In an effort to impress job-seeking scientists with the quality of their intellectual community, Google recently published some of the academic papers written by their search scientists.

This page certainly seems impressive, especially when I read titles like: "A Critical Evaluation of Commensurable Abduction Models for Semantic Interpretation," and "Noisy Time series prediction using a recurrent neural network and grammatical inference."

How will this information be useful to those seeking a higher ranking in Google? I'm not sure yet, I haven't finished reading all the articles ;)

If you learn anything here, please post!

magic_majax · #2 (**permalink**) 07-05-2004, 04:55 PM

Hello Garret,

I have registered in order to talk to you. My intro is here:

http://www.webproworld.com/viewtopic.php?t=23791

Have you seen this?

http://www.mobileread.com/forums/sho...?threadid=1670

It is the google checksum code. I found the link here:

http://www.seroundtable.com/archives/000598.html

I originally began at www.searchguild.com - in their google forum. I found your post via a google search on the "google adsense algorithm" - where it is listed on the first page, no doubt by virtue, at least in part, of the outbound link to one of google's most highly regarded intellectual pages - that page undoubtedly has a PR10.

I am interested in the possible way that google can build up advanced demographics through harvesting data via its AdSense scheme, Toolbar and search engine.

I will try to read as many relevent articles as I can on the page you are linking to. If you have anything to say please pm me for my e-mail and we can discuss this. I am not a scientist but I have a basic grasp of the google algorithm and a fairly good idea of ontology and semantics.

Edited: Those articles are totally beyond me. At least the few I browed through seem to be. I just lucked on an approximation of the google algorithm and have been experimenting with it ever since. Right now I'm not really even trying to use my site to get high rankings - I'm just letting it slide into obscurity. just working on a page here and there.

The next thing I want to write about is google itself.

ronniethedodger · #3 (**permalink**) 07-05-2004, 10:13 PM

Quote:

Originally Posted by magic_majax

I will try to read as many relevent articles as I can on the page you are linking to. If you have anything to say please pm me for my e-mail and we can discuss this. I am not a scientist but I have a basic grasp of the google algorithm and a fairly good idea of ontology and semantics.

Try this for starters Keyword Semantic Connectivity. While it does not deal with Google specifically, it will help you understand the nature of SE's overall and how they operated under the hood.

Dave Hawley · #4 (**permalink**) 07-06-2004, 12:47 AM

I guess my Google job application won't impress them after all :)

Looks like Google has the geekeset geeks.

magic_majax · #5 (**permalink**) 07-06-2004, 01:28 PM

Quote:

Originally Posted by ronniethedodger

Try this for starters Keyword Semantic Connectivity. While it does not deal with Google specifically, it will help you understand the nature of SE's overall and how they operated under the hood.

Yes, thanks Ron, I honestly don't have the scientific background to start at the high-end where those articles kick-off. That looks like a very interesting topic.

Much appreciated.

DrTandem1 · #6 (**permalink**) 07-09-2004, 09:30 PM

This is simply high-tech blither referring to artificial intelligence of sorts. In plain-speak, here is how Google takes simple search terms and references/associates them with what the person is actually trying to find.

With certain rules applied, the search engine will be better at finding relevant web pages. That is, the pages are relevant to what the searcher is trying to find. Using the clues of semantics and grammar, the engine will have not only have a better "understanding" of what the searcher is trying to find, but also what the web page's content means.

When the equations are close to equal, that being what the searcher is trying to find (according to Google) and what pages are delivering the relevant content (again according to Google), the results will be posted in the order of the most relevance and probably balanced with other factors such as links, etc.

All this means is stop trying to out-think Google. That is a waste of time. You should simply be more in tune with your target audience. In other words, configure your site's content to match your audience's expectations. Google will deliver them to you.

amabaie · #7 (**permalink**) 07-09-2004, 10:02 PM

Quote:

In other words, configure your site's content to match your audience's expectations. Google will deliver them to you.

Well spoken. Er...written.

Dave Hawley · #8 (**permalink**) 07-09-2004, 10:07 PM

Most definitely good logical advice.

Deep13 · #9 (**permalink**) 07-09-2004, 11:55 PM

i found some more cooll stuff..

Increasing PageRank through Reinforcement - http://citeseer.ist.psu.edu/629775.html

Link Analysis Ranking - http://citeseer.ist.psu.edu/601420.html

lot of stuff to read :)

Deep

Dave Hawley · #10 (**permalink**) 07-10-2004, 12:01 AM

Deep13, considering the page is about increasing PR and the page has a PR0 I think there's message there :)

Don't Follow The Advise!

Deep13 · #11 (**permalink**) 07-10-2004, 12:34 AM

Quote:

Originally Posted by Dave Hawley

Deep13, considering the page is about increasing PR and the page has a PR0 I think there's message there :)

Don't Follow The Advise!

what about the other link?
i was just searching that site and found those links so thought will share with you people :)

Deep

Leader · #12 (**permalink**) 07-12-2004, 03:48 AM

Quote:

Originally Posted by Dave Hawley

Deep13, considering the page is about increasing PR and the page has a PR0 I think there's message there :)

Don't Follow The Advise!

They have a robots.txt that says

User-agent: *
Disallow: /

Therefore, no PR or listings.

I don't know if the advice is any good, though...