LSI versus Referential Integrity

[SeoGuy]

Hi all, im kinda new here so HI ALL...

There is this new roumor that LSI is actually not strong enough to be used by google..

And that Google would be using referential integrity instead ---> (find it in wikipedia)

As this would be a lot faster to use and a lot more aqurate then LSI

So Gary what have you heard about this?

[kwblue]

These are completely unrelated, so I am confused.

Latent Semantic Indexing relates to linking (although I call it LSL because that's what it really is).

Referential Integrity is all about how a database is structured from primary keys to foreign keys to relationships.

So....... what do you mean here? Is there an article where Google decided to re-define what 'Referential Integrity' actually means?

[GaryTheScubaGuy]

Hi Guys,

My understanding is this;

Referential Integrity relies on a reliable keyword base and breaks if no relative keyword is in place. Sort of like an SQL database. This database is internal and therefore works only if the data sets being analysed contain and references relative terms to the query. Its like taking relative terms directly from a thesaurus.

LSA (Latent Semantic Analysis) relies on associative words whether they are available or not and relies on analysed data/content on the Web. Rather than a Thesaurus, this is like bunching terms together that may not always be referenced together.

I also use the example of a truck. In the US its called a truck, in the UK a Lorry, but int the New Zealand its referred to as "ute". All are different names for 'truck'.

In each instance these words could have other meanings;
Truck - (nautical) a circular or square piece of wood fixed on the head of a mast or the top of a flagstaff
Ute - a member of an American Indian people of Utah and W Colorado
Lory - any of several small, usually brilliantly colored Australasian parrots

So in the instance of Ute, the results from a LSA-heavy algorithm would return all sorts of variations. If you do a Google search you will see that the actual term we are referencing here doesn't show up (Google.co.uk results) until the 10th spot. These results are based on an algorithm that relies heavily on back links and page titles to determine relevancy.

Doing the same search but using a tilde before the term (this is a utility Google has furnished to play with - ~ute) produces these results. If you look at the first listing its for a website that doesn't even use 'Ute' in its term.

In fact, when you search the whole website it doesn't even use 'Ute' once. The curious thing is that without the tilde sign before the term, this website isn't in the top 100 results.

Now throw in the added element for regional searches and language-based searches (searching for 'Ute' in New Zealand and using language-based filters, or searching the same but using the regional-based filters) and you see the soft-market for websites that use LSA when writing content and acquiring back links. (and additionally using Advanced Geotargeting), and you can see that this really does get complicated.

So although both operate on the same theory, they both use unique data.

kwblue

LSA (Latent Semantic Analysis) that drives LSI (Latent Semantic Indexing) is meant to be a part of an algorithm that determines associative terms (depending on the level at which it is 'turned up') to weigh web pages and websites based on the relativity or theme base as determined by the search query.

When you apply this to link building, it IS as you say a factor involved in link building, and is an extremely beneficial technique of altering the anchor text in your internal and external (pointing inwards) links when optimising for SEO.

Utilising LSA in your content, theming and silo'ing of your site (via internal links) is an advanced technique that will put you head and shoulders above the rest.

Utilising LSA in your back links is a super-charged SEO strategy, especially when targeting niche and long tail, and more specifically when trying to give your homepage and Hollywood terms a boost into the top 10.

But that's just my personal understanding of the difference.

Gary