LSI Explained
- SEO |
Google has lots of cash for sure and can buy any technology that exists but the reality even in 2011 is that the technology still is a long way off even the most basic "humalinke" understanding of even the written word by any so called AI or "artificial intelligence".
What Google has done is to invest heavily into "LSI" or "Latent Semantic Indexing" which is simply a mathematical way of identifying related words which are then weighted due to proximity to each other.
This is why they bought companies specialising in LSI such as "Applied Semantics" and "Metaweb" amongst a few others like it for example
Sounds a bit gobbledy gook?
It's not so hard to understand really it just looks for words it has in its dB and finds others close to those words and works out how closely they relate to each other "semantically"...
If it finds "dentist" and three or four words later it also finds "teeth" then semantically it scores it highly especially if, in the same "web space", it also found a H1 or H2 tag that said something like "Tooth Whitening" and maybe the site itself was called something that included "dentistry" in the name etc...
It can then determine and associate correctly the word "ford" for example on a car site because it finds other words like "motor" and "pickup" near by as opposed to "ford" found on a nature related site where it finds words like "stream" or "river" nearby.
This technology allows it to confirm the appropriate niche to store a web page under...
It adds up all the scores and "knows" that it was from "similar" paragraphs in the same blog post (Google likes blogs and it loves in paragraph LSI contextual links!)...
It then "asks" itself do I trust this place I found this content...?
If the answer is yes it then scores it appropriately and decides if it is
a) interesting enough to index and
b) interesting enough to take up disk space and be cached...
The bigger the total score the better value the link has back to whatever site...
What it doesn't do and cannot do is actually "read" any text and "understand" it!
This means that for backlinking purposes it is very important to be "LSI correct" but not necessarily "keyphrase" perfect... a VERY important distinction missed and misunderstood by many...