Call for five SEOs

[orion]

CALL FOR FIVE SEOs

EXPERIMENT #1: ONLINE DISCOVERY OF SECONDARY TERMS ASSOCIATED TO A THEME

Are you an SEO specialist looking for keywords relevant to a theme? Are you familiar with c-index calculations? Would you be interested in participating in a controlled experiment on keyword semantics? If so, this may interest you.

We are conducting the first of several online experiments on semantics and are looking for five (5) volunteers.

This first experiment pretends to discover secondary keywords associated to a theme represented by an initial key phrase. Participants can keep and use the results of the experiment at will, in their sites or clients' sites. No cost. No strings attached.

Conditions

1. Each participant should submit one key phrase consisting of two terms. Submissions should be received no later than the midnight of 09-07-2004 by private or regular email. We reserve the rights to accept or reject participations.

2. All candidate phrases must be in English, delimited only by a space (no hyphens or other delimiters). Example: discount hotel

3. Phrases should include a c-index value equal or greater than 25 parts per thousands (ppt) when searched in Google using its default query mode (FIND ALL).

Why wait? Start submitting, now.


Useful Notes

c12-index = (n12/(n1 + n2 - n12))*1000

where n1 = #results for k1, n2 = results for k2, n12 = results for k12, and k12 is a phrase consisting of k1 and k2 terms.

The 25 ppt mark was obtained from a test pool of many popular phrases queried in Google. We intent to run additional experiments without this requirement as well as with retrieved set of results of variable size.

The theory behind c-indices and this experiment is described in the Keywords Co-occurence thread (http://forums.searchenginewatch.com/showthread.php?p=12122#post12122)


Orion

[Nacho]

Thank you Orion! I have sent you my keywords via email. I wish you sucess with your experiment Amigo. :)

[orion]

Hi, Nacho

More than happy to include you. I need a candidate phrase with a c-index equal or greater than 25 ppt in Google.


Orion

[AussieWebmaster]

I will be submitting also once I calculate which terms apply.

[orion]

Hi, Aussie

It's an honor to have your participation. Just PM or email me a competitive 2-term phrase including its c-index (equal or greater than 25 ppt) in Google.

Orion

[Joseph Morin]

orion,

Count me in, I work on enterprise level sites with a lot of analytical software that might help in your research. I'll also determine which terms might be most useful for your research. If you have any guidelines or preferences, please let me know so that I may pick an appropriate industry for example.

[orion]

You're in. Just send me by private or regular email a two word phrase with a c-index equal or greater than 25 ppt in Google. See conditions above.

Orion

[orion]

Hi, Chris_D. I got your PM. You're in.

Orion

[Marcia]

Darn, I'm a regular keyword generating machine and in the process now with a burgeoning list for a new site with a two-word theme that is a HOT white hat type consumer item. But I never have gotten into c-index computations; it's hard to believe it could out-do women's intuition.

Too bad, I'd gladly participate to see if a man with a machine can out-do a woman armed with just a brain. ;)

[Jill Whalen]

That c-index stuff is really greek to me. Perhaps someone could write a c-index calculator so that people could easily figure out if their phrases qualify?

Am I the only one who hasn't the foggiest idea of what it all means?

[Nacho]

Jill,

Take a look at your email. I just sent you an excel version on how to calculate a c-index. It's not that hard, really, now that Orion has shown the formula.

Enjoy :)

[Jill Whalen]

Any Excel doc is usually greek to me too... :(

[AussieWebmaster]

Any Excel doc is usually greek to me too... :(

Guess the Olympics have you seeing Greek everywhere!

[Nacho]

Here is the calculation of the c-index in Excel:

http://www.ihispanic.com/sewf/c-index.gif

Where column B has the number of documents found on the Google index for that keyword.
http://www.ihispanic.com/sewf/serps-c-index.gif

[rustybrick]

What so Greek about this page (http://www.miislita.com/semantics/c-index-2.html), :eek:

http://www.miislita.com/searchito/images-marketing/co-occurrence-0.gif

http://www.miislita.com/searchito/images-marketing/co-occurrence-7.gif
good thing we are doing a single co-occurance here. :)

Beutiful resource Orion, much easier reading your summary (http://www.miislita.com/semantics/c-index-1.html), thank you.

[Joseph Morin]

Since this series of articles are intended for average readers, SEOs, SEMs and others, the following information is presented in non technical terms.

...I'd hate to see what he calls technical (!!!) :eek:

[rustybrick]

To tell you the truth, I love this math. Reminds me of college.

[orion]

Hi, Jill and Marcia. Just PM a phrase with c-index.

Hi, Nacho. Yeah. With an Excell document, c-index calculations are straightforward.

Hi, Rusty and Chris_D. I got your PMs. You're in. Excellent, competitive terms.


Orion

[orion]

1. Sample calculations with a hand calculator

I'm using Nacho's excellent example. k's are terms and n's are results. The query is mexican food. Google is queried in default mode (FINDALL)

k1=mexican n1=10,400,000
k2=food n2=124,000,000
k12=mexican food n12=3,420,000

c12 index = n12/(n1 + n2 - n12)=0.02611

Multiply this result to express it in parts per thousand (ppt)

c12 = 26.11 ppt

A c index is a measure of the co-occurrance of terms in a queried database. Due to the large volume of documents, these are small values, thus are multiplied by a factor --1000 in this case. Since SEs are constantly updating their index, c values may change in time. Still, trends can be extracted from time series analyses.

2. Confidentiality

Oops, my fault. I forget to mention this part in the original post.

a. In exchange for their participation, volunteers will receive their raw data of discovered terms. I'm not planning in publishing participants raw data.

b. However, without identifying volunteers, I'm planning to post at the SEW thread some partial results (top 5-7 discovered terms) of each trial, followed with a brief explanation. See recent posts at the Keywords Co-occurrence thread for a sample.

c. I'm planning to present partial results and a detailed theoretical explanation of the exp at my site.

Feel free to opt out b and c. If someone feel his/her partial results should not to be included in b or c or both, please let me know now.

Orion

[orion]

1. Last call. Let me know who don't want their partial results published. I'll provide credits to all volunteers by name but will not identify who submitted what, unless requested.

2. Someone asked me about the 12 subscript. This is to distinguish a k1+k2 phrase/c-index from its transpose k2+k1 phrase/c-index.

3. The same person asked me about why the use of the c-index mark of 25 ppt (parts per thousand) in this experiment.

This mark is the result of testing many two-word phrases in the Google database, thus is valid for experiments conducted in Google. The mark serves as a reference point, no more, no less and to insure some degree of semantic connectivity between candidate terms. Why? Terms co-occurring frequently are often semantically connected in some way. For example, many think in Hawaii when they hear the term "Aloha". This is a semantic association.

Now, let's consider the word k1=aloha and some candidates k2 terms; ie., california, hawaii and florida. A quick FINDALL search in Google reveals that

k1=aloha n1=2050000; k2=california n2=120000000; k12=aloha california n12=307000 || c12=2.52ppt
k1=aloha n1=2050000; k2=hawaii n2=28800000; k12=aloha hawaii n12=845000 || c12=28.16ppt
k1=aloha n1=2050000; k2=florida n2=63900000; k12=aloha florida n12=204000 || c12=3.10ppt

Clearly the c-index results reveal the term aloha is more semantically connected to hawaii than to california or florida, with common usage dictating this association. The above mark is used in this experiment as a generic reference point from which one can derive secondary terms that may reinforce a theme. The idea is to start with a valid ground from which secondary terms are identified.

On the other hand, these results reveal c-indices can be used to identify terms not semantically associated between themselves or with a theme.

Orion

[Alan Perkins]

One English key phrase I checked has a c12 of 43.30 ppt on google.com, but the same phrase has a c12 of 357.82 ppt on google.co.uk (UK results only). Clearly locale can influence semantic association even within the same language. :)

BTW this a highly competitive phrase with n1 and n2 in the hundreds of millions...

I have PMed you the phrase, orion.

[orion]

Hi, Alan

I got your PM. Excellent

Alan. Well put and very true. In post #9 of the "Keywords co-occurrence..." thread and in subsequent posts I discussed that c-indices are excellent tools for discovering and targeting the right terms, especially with regional directories. This is because semantic association is culturally-driven. We use the tool for clients interested in targeting specific markets, demographics and regions. It is an excellent tool for targeting vertical portals and topic-specific directories. However, as any tool, in the wrong hands or with the wrong knowledge it might help nothing. c-indices, ef-ratios and other tools I may present are not silver bullets.

Kind suggestion: let's take c-index discussions to the Co-occurrence thread. As a thread for beta testing stuff this thread may be running out by midnight.

Orion

[orion]

Thank you all for participating in this experiment on semantics. Your participation and feedback was instrumental in making the exp a success.

Results have been published and an SEW thread for its discussion is available at http://forums.searchenginewatch.com/showthread.php?t=2031


Orion