homepage Welcome to WebmasterWorld Guest from 54.224.179.98
register, free tools, login, search, subscribe, help, library, announcements, recent posts, open posts,
Subscribe to WebmasterWorld
Home / Forums Index / Marketing and Biz Dev / Link Development
Forum Library, Charter, Moderators: martinibuster

Link Development Forum

    
Determining Link Context: Topic and Themes
Brett_Tabke




msg:424214
 9:42 pm on Jun 26, 2000 (gmt 0)

I've had a fair number of questions today about the tool I mentioned at the U of Toronto called Topic [cs.toronto.edu].

Topic was developed by the U to study link context and web page reputation. It shows just how easy it is for the search engines to determine what your website is about from a few simple criteria.

What it does:
a) does a search on altavista for "links:yourdomain".
b) it looks at up 500 returned results.
c) it does NOT download those pages. It only connects to altavista.
d) it takes the search results and indexes the results.
e) it throws out the stop words found in the results.
f) by indexing the results it can come up with the top 10'ish primary keywords found in page titles and descriptions leading back to your site.
f) it will also download the page at the url you entered, and index
it only for display comparisons.

From that info, it will display a short word list. For example, lets look at what Topic returns for Search Engine World.com:

Query : searchengineworld.com
Loading pages from Alta Vista 5604 terms extracted from 347 incoming links examined ( out of 604 links available) 46 links were found duplicates and were removed

The page is known for:
( 6.20857e-05 ) Download
( 5.15886e-05 ) Search engines
( 6.83832e-06 ) mp3
( 1.92028e-06 ) guide
( 1.73235e-06 ) Discount
( 1.58995e-06 ) Promotion
( 1.53808e-06 ) Jewelry
( 1.28865e-06 ) vacation
( 1.24408e-06 ) Free
( 8.83957e-07 ) traffic
( 7.30114e-07 ) Tips
( 6.79685e-07 ) Friends
( 5.63267e-07 ) domain
( 3.81656e-07 ) Website
( 3.64114e-07 ) Tools
( 3.47925e-07 ) designer
( 3.47352e-07 ) experience
( 3.34078e-07 ) Categories
( 3.32846e-07 ) Rental

The page content:
( 9.17414e-06 ) engines
( 9.88704e-07 ) Search Engine
( 6.28938e-07 ) traffic
( 3.20817e-07 ) Tips
( 1.9782e-07 ) Tools

The returned scientific numbers show you how each word relates to each other and should be used only as a gauge.

Notice some of the stray words that don't appear to be ontopic with search engine world? Jewelry, Mp3, vacation, rental?! That is what Buddy Links was doing to us.

Other sites of ours not in Buddy Links show a set of words right on topic (theme).

That is how easy it is for a search engine to perform a context links check on a shoe string budget. Easy peasy. It is one of the primary reasons I concluded Buddy Links.

As I said in the email, I've been working on a similar tool that should be online later in the week.

Brett


 

NFFC




msg:424215
 10:38 pm on Jun 26, 2000 (gmt 0)

No doubt a great tool, this will help anyone focus on the theme of their site, already useful for me.

Having said that www.yahoo.com is rather amusing.

thewebguy




msg:424216
 10:57 pm on Jun 26, 2000 (gmt 0)

Love the link I have used it with severeal site already and revamping has begun. Thanks Brett!!!!
NFFC why find yahoo amusing did you run it through this program?

pshea




msg:424217
 12:30 am on Jun 27, 2000 (gmt 0)

So, what I'm getting now with all of this, is that we have to carefully consider the link location page of those sites we court for links. That "themes" is actually much more narrow in this context than in regular usage of the word. It's down to almost exact word matching of the linking page title and description.

The next question lies in those scientific numbers, how well does this method understand which words are in the same theme to the user's search string. That is that "jewelry" and "bracelet", while not an exact word match to each other are in fact part of the same theme.

If everyone throws my link on a page they name, "links", however well we are matched themeatically to the actual site overall, unless that page can, in its other page title words and description, overcome the presence of the word "links", then that link could hurt me more than help me and I should just keep looking. That sounds quite severe but if I'm starting to get the fallout of your discovery, I think that's the plan.

NFFC




msg:424218
 8:03 am on Jun 27, 2000 (gmt 0)

thewebguy,

Yes just run www.yahoo.com through the program and see what yahoo is known for. A result of all those "exit if under 18" links which really shows the accuracy of the tool.

thewebguy




msg:424219
 1:09 pm on Jun 27, 2000 (gmt 0)

I just ran it for Yahoo also..thats funny.
How long for the SE to pick up your theme?

Brett_Tabke




msg:424220
 1:24 pm on Jun 27, 2000 (gmt 0)

Different for each one. Inktomi picked up on ours last fall and delivered quality keyword referrals since then. Not a killer ranking in the bunch, but a broader range of kw's.

Alta picked it up immediatly. No questions asked. The problem with Alta, is it picked up the wrong one. To this day, search engine world pulls from exotic car engine keywords.

thewebguy




msg:424221
 1:36 pm on Jun 27, 2000 (gmt 0)

I ran the topic report on one of my sites and nailed it on the head. but site is not even ranked?
I then I ran it for top sites and not even close to theme? I dont understand that?

Brett_Tabke




msg:424222
 1:38 pm on Jun 27, 2000 (gmt 0)

That one is using Alta. There is limited evidence that linkage plays a huge roll in Alta. At best it is 1-5% of the algo.

If you have a similar tool, (and you will have soon), we are trying the same with Inktomi - which is leading to breath taking results.

thewebguy




msg:424223
 1:48 pm on Jun 27, 2000 (gmt 0)

I am taling about AV. Ran the top site in AV for my search word and seems off the theme?
Also would it effect if you have like 50 domian names and make sites for them all and link them all together (with a 1pix hidden link) would this help with links and themes shouldnt it?

Edited by: thewebguy

Brett_Tabke




msg:424224
 1:55 pm on Jun 27, 2000 (gmt 0)

The topic tool is more about links back to your site. It is just a part of the theme algo, but does represent how easy it is for engines to do.

LinkMe




msg:424225
 12:49 pm on Jun 30, 2000 (gmt 0)

Hi all. My first post here.

"That "themes" is actually much more narrow in this context than in regular usage of the word. It's down to almost exact word matching of the linking page title and description."

It is important to remember that this tool is only an example of what can be done with limited information. The engines have access to more than just the title of sites that link to you. They know the linked text itself and the words near the link. With this additional information their relationships can be much more complex. Or maybe simpler.

Bradley




msg:424226
 5:41 pm on Jul 1, 2000 (gmt 0)

Let's say you have a site that is very general in nature (i.e., a search engine/directory) and you want to implement themes for SEO. Would it be beneficial to break up your site into multiple subdomains? For example, lets say I own search.com (Wouldn' that be nice!)

If I wanted to be focused and target themes, would the search engines indexing my site give me credit if I created sites like:

cars.search.com
movies.search.com
shirts.search.com

Do the search engines think these are totally unqiue websites in and of themselves, or do they realize they are all a part of search.com? Do you think it is beneficial, in the first place, to create these type off subdomains off of your main website? What are the advantages/disadvantages?

I look forward to your reply.

rcjordan




msg:424227
 6:00 pm on Jul 1, 2000 (gmt 0)

>Do the search engines think these are totally unqiue websites in and of themselves

yes. overall, it is to your advantage to used third level domains.

Brett_Tabke




msg:424228
 11:56 am on Jul 3, 2000 (gmt 0)

Ya, they'll see thrid level subdomains as unique. Certainly.

Global Options:
 top home search open messages active posts  
 

Home / Forums Index / Marketing and Biz Dev / Link Development
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved