It is easy to become confused with so many different acronyms being used for techniques to help improve SEO, so you could be easily forgiven for not knowing exactly what LSI keywords are or how Google uses them.
LSI stands for Latent Semantic Indexing, and Google actively looks for keywords or phrases that are semantically related to a particular topic. So for example, if you were talking about laptops, then associated LSI keywords could include hard drives, memory cards, wireless mouse, laptop carry cases, power cables and USB chargers.
However, according to a top Google engineer, LSI keywords are not used by Google. Confused? It’s not surprising!
So what exactly is Latent Semantic Indexing?
Latent Semantic Indexing, also known as Latent Semantic Analysis, is actually quite an old keyword technique developed back in the 1980s. It was based on a mathematical formula to help deliver related information that has been indexed, but not under the keywords that a searcher uses.
So for example, if you were looking for ‘laptop sales’ Google would deliver results for similar synonyms – words or phrases that either means the same or nearly the same as the keywords you use to search with.
However, there is so much diversity used with keyword searches, especially when people use different words to describe the same product, service or idea that results can vary wildly. So if you have two almost identical websites offering the same product or service, your results would show the website in your search results that contain the keywords that most closely match the words you used to search with.
While this isn’t a bad thing because both of these websites would be relevant to your search, it means that the one you get at the top of your search results will more closely match your search words and phrases.
Where things get interesting is with the use of polysemic keyword searches. Polysemic words and phrases are those with many different meanings. So if you wanted to search for a mouse to use with your laptop, if you Google the word ‘mouse’ the chances are you will get results relating to both computers and rodents.
Basic search engines that don’t understand the different meanings of polysemic words and phrases are more likely to deliver you with useless results. They are not capable of understanding words or phrases in context as a human does.
How does LSI help?
Google and other major search engines employ LSI techniques to overcome these issues. It involves using complicated mathematical formulas to establish relationships between keywords and key phrases drawn from a set of documents.
So if you run LSI using a set of documents about laptops, then the search engines will be able to figure out the differences between search results for a computer mouse compared to a rodent. This allows them to go far beyond simple exact keyword matching to be able to deliver more relevant results within the context that you are using your keywords.
Does Google use LSI?
LSI goes a long way to solve the problem of keyword context and meaning, so you can understand why everyone believes that Google and the other major search engines must use it to enable them to deliver such good search results.
Without LSI, all the search engines would only be capable of delivering exact keyword matching results, with a lot of references to cute mice with whiskers instead of computer hardware.
However, because LSI technology was developed in the 1980s a long time before the invention of the internet, it was never intended to be used in search engines. It simply wasn’t engineered to be used with such large sets of documents.
This is why Google has spent years developing better technology to solve the same problems that LSI does, but using a world vector approach that scales much better and works on the internet.
While LSI was created to index already existing document collections, the internet is so much larger and is a dynamic system – not a static system.
Many people say that without the invention of LSI, Google and other search engines would never have been able to evolve as far as they have.
LSI was a breakthrough technology for its time, but because there are billions of pages in Google’s index and these pages are constantly changing, LSI wouldn’t work for them because it was only created to index static document collections that don’t change.