Why You Need To Understand Latent Semantic Indexing

By Author: Larry Tang
Total Articles: 512
Comment this article

Latent Semantic Indexing (LSI) is the underlying root of the entire major search engines' algorithms. There is a complicated Wikipedia version here

(http://en.wikipedia.org/wiki/Latent_semantic_indexing) ...

that goes into detail of the Matrix relationship of words, but to really understand it you will need some knowledge and understanding of Calculus and Advanced Algebra, and have not slept through the Derivatives and Matrices parts of class. I will attempt to show you a simplified version of it.

Guess the theme/keyword of this next sentence:

"I love linguini, spaghetti, and angel hair. I also like rotini, bow ties, and macaroni."

Here is the same concept of the sentence with stuffed keywords:

"I love pasta. All kinds of pasta. Without pasta, I couldn't survive. There isn't a type of pasta that I don't like."

Also you can see both sentences are related to "pasta", the former you can read as a regular human being that they person likes pasta, the search engines will see the same thing. The second is almost unbearable to read, and as a result of them keyword ...
... stuffing the search engines will flag this as spam/keyword stuffing.

All the search engines determine the way words relate to one another use LSI. Now each search engine's matrix of word relationships is different, but their whole underlying theme is to deliver more accurate search engine results.

In the example above, most search engines will give a higher ranking to the first example, and correlate the page with the word "pasta". In the second example the search engines will correlate the page with nothing, not even pasta. They would disable the word since it is seen as keyword stuffing. 18% of the words in the second example are "pasta", that's too high. My recommendation is to keep all words under 8%.

Now that you understand what Latent Semantic Indexing is, you will have a huge advantage over the competition. The search engines do not want to deliver pages considered spam, so instead, use synonyms and related topics when creating content.

LSI is the backbone of all search engine algorithms. The reason being is that each search engine needs to decode and decipher what each page "means", and what the content is talking about. The only way to do that is to find correlating words on the page, and then put the "page" into the search engine's proper category or index.

The benefit of this is that people talk and write naturally more like example #1, than they do in example #2. So when creating content, write like you are writing for your visitor and provide examples and images as visual references. This will give you the benefit of having visitors enjoy reading and viewing your content, and give you a huge leg up on the competition that may not know what LSI is and how to properly use it.

There is a cool tool that allows you to see in 3d graph format how a website and/or keyword relate to one another. It is called TouchGraph GoogleBrowser V1.01 (http://www.touchgraph.com/TGGoogleBrowser.html). It is very useful, and can helped you in finding niches that Google believes are related to your industry.