Does Latent Semantic Analysis affect Google results?

[j10222]

Does Latent Semantic Analysis play any part in the Google algorithm?

ie: if searching for "~widgets" I get keywords such as "widgetitus" and "Wideye" associated with "widgets", will it hinder my results not integrating these words in close proximity to my main keyword?

Hope this makes sense, not sure "widgets" is the best example!

[tedster]

Google doesn't use standard Latent Semantic indexing - one reason is that it requires too much computational overhead. However, they have their own parallel form of semantic analysis - see the thread Phrase Based Indexing and Retrieval [webmasterworld.com] for a discussion of Google's six patents in this area.

Side note - the author of these patents is Anna Lynn Patterson. She left Google and is now one of the co-founders of Cuil.

Don't expect an easy read from these patents, or a quick answer to your question. A very top-level answer would be that without at least some common co-occuring phrases, you may have more trouble ranking. However, with too many, you can be penalized.

And don't depend very much on the public information you can see with the tilde operator. There are some very peculiar oddities in there, simply based on coincidence and not on word meanings.

These six patents are pretty much there to protect the way in which Google reduces the computational overhead - you'll see rather quickly that reverse engineering the calculations is not for anyon without some kind of super-computer and massive spidering capabilities.

[BroadProspect]

it does in the indexing of the site part, a site can be retrieved from the index for the word "automobile" even if the word is not on the page or in the anchor text pointing to it, and the page is talking abour a "car"

/BP

[j10222]

Thanks for both comments, am I right in thinking then BP that It would affect results for "car" if I also included "automobile" in the main copy.