In this article I am going to discuss elements of Latent Semantic Indexing (LSI) technology that you need to be aware of.
As I discuss this I am probably going to upset some of the SEO mathematicians out there, so I will go ahead and address the fact that I am not talking about the “mathematical” implementations of LSI, I am talking about the marketing aspects. I am also not talking about LSI as a whole, I am talking about certain elements of LSI that Google is currently using as a factor to determine the relevance of a web page and overall website. When I say LSI, I am merely using it as a “catch all phrase” for what I am about to discuss as it relates to the marketing world (not mathematical).
Google has had a long struggle with search engine spam from the black hat and grey hat webmaster community. Why play along with Google they say? These webmasters actually believe they are smarter than the collective knowledge of all the engineers, mathematicians and programmers at Google… Well okay.
Anyways, we have come to discover that it actually pays to play along and not in a small way. Myself and my colleagues are quietly and systematically swallowing our entire vertical markets and Google is loving it.
We are seeing top 3 rankings for competitive terms within 90 days. We are seeing almost immediate traffic (1 week or less) for the longer tail keywords contained within our website content.
We are discovering that something has leveled the organic search playing field for us. Why are we outranking the big dogs?
Latent Semantic Indexing (LSI)
The element of LSI that I am referring to, in particular, is the existence of what we call the “Co-occurrence Matrix”. I can hear the mathematicians already!
The co-occurrence matrix is simple to understand. Without getting into mathematics, it is a way for Google to understand relevance based on what “humans” find relevant.
For example:
Let’s say you are an expert on Tiger Woods and decide to write an article about him. Your article will naturally include words that are associated with Tiger Woods like golf, Nike, Gillette, EA Sports, Golf Digest, Target, WGC Bridgestone Invitational, etc.
As an expert you know how Tiger Woods is associated with all these words. Because you actually know what you are talking about you will naturally include the other words which, seen from a novice, seem to have nothing to do with Tiger Woods at all.
To continue the example lets say that there are other experts, like you, that also know what they are talking about and the content of their website contains all sorts of keywords that they associate with Tiger Woods.
Google realizes that “trusted” sites “know” what they are talking about. Trust is measured by incoming links from other “trusted” sources. The higher your Page Rank the more Google trusts what you write about.
To explain this in simple terms, Google takes a snapshot of all the “highly trusted” web pages that have the phrase “Tiger Woods” on them. Google then looks to see what other keywords they are associating with Tiger Woods.
Because Google trusts these experts, based on their Page Rank, they add their keywords to what we call the “co-occurrence matrix”. The co-occurrence matrix is Google’s hypothetical index of what keywords are commonly found “IN ASSOCIATION” with the main theme of a web page.
I say this is hypothetical because Google would not dare to admit it. Testing by myself and my colleagues reveals that this is fact. The co-occurrence matrix does exist. Several Google Patent Applications also reveal that they intended to move in this direction.
In the case of the example above, when Google indexes your new article and sees the word “Tiger Woods” as what you are attempting to rank for, they then will cross reference your content to see if you also discuss any of the key phrases that have been established by “trusted” sources.
If your content fails to include these words and phrases then you will not rank for the phrase “Tiger Woods”.
If your page contains this “expert verbiage” then you have just successfully leveled the playing field in your market because 99.9% of webmasters have no clue that this is even going on.
Now, if I were a spammer or a black hat webmaster I would try to rank for the phrase Tiger Woods by doing a keyword density analysis of the top 10 websites to see how many times they include the phrase “Tiger Woods”. I would then throw up some garbage content to try to mimic that keyword density to achieve rank.
This strategy used to work, but not any longer. The days of keyword density analysis are over. The age of “Theme Density” analysis has arrived.
Webmasters are struggling because their market share is dropping fast. Google has begun their crawl using this technology. If you don’t take the time to learn about it then you will be left in the dust.
How to Level the Playing Field
My students at University 20/20 have been learning to use a proprietary system where we are able to extract contextual data from the top trusted sites and analyze that data to determine exactly what we need to write about in relation to the keyword we want to rank for.
Sprinkle a few back links to a newly created, “theme dense” page and we are seeing top 3 rankings for keywords that were previously thought impossible to own within a week to 90 days. They were previously thought impossible because of competition. The competition has no clue about this.
Currently there is only one tool that automates this process but you can do this research manually. I always advise my students to learn the process manually so that they know what they are doing when they use a tool. This also helps to sharpen their knowledge and it keeps us all on the cutting edge at all times.
You need to take the time to learn about this because we know that Google is using it and the other major players like Yahoo and MSN are quickly trying to follow suit in order to hold on to their market share.
LSI has made it possible for an automated search engine algorithm to come as close as possible to mimicking what the human mind finds relevant to their search queries. Google knows if you are trying to fake your way up the ranking chain by consulting with the “experts” in every market… AUTOMATICALLY!
In the next article I will discuss more about “Theme Density”, LSI and the co-occurrence matrix.
Until next time,
Charles Heflin – Professor
http://www.university2020.com
Digg 
