Friday, 8 December 2006

Latent Semantic Indexing

This is a strange one; there has been a lot of talk lately in Search Engine Optimisation (SEO) circles regarding Latent Semantic Indexing (LSI). This is nothing new, there is always something new to talk about. But what I’m concerned about is the almost contradictory elements involved in the whole process. A very brief summing up of LSI - it is a technique of analysing the occurrence of terms in a document – in my field of work the document means the page of a Website.

Ok that seems quite simple, and in this indexing of terms (language-semantics) we find out that the terms or words are broken down to their basic (lowest denominator word) element; so for instance the word ‘edited’ or ‘editor’ would be broken down to the word ‘edit’. This is where it becomes a little tricky because as we know the words ‘edited’ and ‘editor’ mean quite different things to us (one is the final accomplishment of the other).

So how are we meant to integrate this form of semantics into our ‘content’ within our Website; many SEO agencies will inform you that across each page of your Website you should vary your keywords: i.e. flower, flowers, flowering, flowery (I didn’t know if that one would spell right), but I ask why? If the search engines are busily breaking down our words to their basic elements what is the point of spending heinous hours studying the art of the verbose.