Sometimes when you are trying to explain a simple idea, people zone out because they think they already know it all. That’s when you whip out a phrase like Latent Semantic Indexing.
It’s a great phrase because it sounds like a super complicated scientific method for… well, something. In reality, it’s sort of what good search optimizers have been talking about for a while.
Latent Sematic Indexing can have a very technical sounding definition:
Latent Semantic Indexing (LSI) is an indexing and retrieval method that uses a mathematical technique called Singular Value Decomposition (SVD) to identify patterns in the relationships between the terms and concepts contained in an unstructured collection of text.
I usually only throw out LSI when I’m losing people while talking about simple semantic document structure as a “best practice”. When I explain the the title tag should encompass the main thrust of the content of the page, it seems almost silly to say that out loud to a room full of adults – but sometimes titles and headlines seem barely thought through.
When I explain that the header tags should be used in a nested fashion people roll their eyes, until I show them they their sites are doing the opposite. Header tags are not a style, they are a semantic queue about what is important on your site.
I actually use a cocktail approach to semantically related phrases and real LSI. If you have a page about Sesame Street, you wouldn’t just stuff it with the phrase “sesame street”, you also want to include oscar the grouch, elmo and big bird.
Learning what words Google thinks are semantically related can help you. Try using the tilde operator, for example: ~bbq and you get a slew of relationships: barbeque, bbq grilling, grilling, bar-b-que, grilled.
What that would tell me is that if I had a website for the best bar-b-que in downtown Raleigh, I would try to get inbound links around all of those semantically related phrases.
You may have heard people talk about a site, or a page being “over-optimized”. Usually in those cases the content is focused on a single term that ends up looking spammy and forced rather than a semantic mix of phrases.
Photo by: pshab