LSI keywords – what are they and do they have a real impact on SEO optimization?
LSI keywords are extremely important phrases in positioning. But does Google really work that way? Check out the answer in the article below.
What is the concept of LSI?
Latent Semantic Indexing (LSI), which can be translated to mean “hidden semantic indexing”, is the process of semantic analysis of a website that search engines use to accurately understand the content of a given website.
In the past, search engines placed great emphasis on the presence of specific phrases in the text – that is, keywords – in order to determine the content of the page. For this reason, it was desirable to have as many occurrences of a given word cluster as possible, which would signal what the content is about. This method can be unequivocally considered obsolete.
Today, Google (as well as every other major search engine) uses a number of advanced algorithms based on natural language processing to discover the general theme of a page. What is it for?
It’s easier to find information on any topic
First of all, if we limit ourselves only to searching for pages by a given key phrase, we would have to know it precisely to find a given page. Information is often cataloged in the search engine using different words than those that a potential customer will use to find such a page.
Example? Let’s assume that we buy catalysts and we will set these two words as the key phrase for our website. If a customer has such a part for sale, but instead enters “car parts purchase” into the search engine, he will miss our website.
This problem seems to be quite easy to solve – it is enough to “saturate” our website with more key phrases. Unfortunately, natural language is a much more complex creation.
Solving the problems of synonymous, polysemic, homonymous words…
Language offers a real wealth of ways to express a concept in different ways.
It is very rich in synonyms, polysemous words and homonyms. A synonym for the word “car” is “vehicle”, “lock” is a polysemism meaning a structure, a part used to close a door or fasten an outer garment, and “wounded” is an example of a homonym meaning someone injured or someone getting up at dawn.
As you can see, it’s easy to misunderstand. Therefore, it is no longer enough to encode the pool of available meanings for a given word in the search engine – it is also necessary to understand the broader context.
After doing a little experiment on your own, you’ll see that Google understands the concept of synonym and polysemics.
Try typing “mouse” into the search engine and you’ll see that in addition to any rodent data, you’ll also see a “See Also:” box in case you’re referring to a computer mouse.
Is latent semantic indexing the answer to all linguistic nuances?
As you can see, Google efficiently manages to tame all the richness of natural language. The thing is… it doesn’t do it with LSI.
John Mu, webmaster trends analysist at Google, says about it directly in a Twitter post: “There’s no such thing as LSI keywords — anyone who’s telling you otherwise is mistaken, sorry”. Although LSI is a real information technology, it was created in the 80’s, before the era of the universal Internet. In addition, the patent on it expired in 2008. Google uses search algorithms, so-called word co-occurrence and bipartite graph co-clustering, which for the vast majority of us will probably be completely unclear.
Anyway, let’s answer the important question: LSI or not LSI, can using related words or phrases improve the performance of our site? The answer is: absolutely yes!
How to find LSI keywords?
For starters, it’s probably worth looking at what Google itself tells us about its search process:
“Understanding the meaning of your search terms is critical to the quality of your responses. (…) In addition to keyword matching, algorithms try to determine how effectively potential search results respond to the user’s needs. When you search for “dogs”, you probably don’t expect a page where the word is repeated hundreds of times. We try to guess whether a given page contains an answer to your query and is not just a copy of it. So search engine algorithms analyze the relevance of the content of the pages – for example, whether they contain pictures of dogs, videos about them, or even a list of breeds. We even consider whether the language of the page matches the language of the request to prioritize pages in your preferred language.”
Google analyzes pages for possible development of the topic, and for this purpose uses pages that contain semantically similar content.
Therefore, the more cross-sectional materials we post on our website, the greater the chance that we will appear high in search results for terms that are similar in meaning. That is: cross-sectional texts, containing a semantically large cross-section of a given issue = higher results. Maybe not directly, but definitely significant. So how do you find related phrases?
Autocomplete search results
Type your phrase into Google and see what ending it suggests. These suggestions indicate that the phrases have already been searched for – which means that you should consider them as a way to enrich your texts.
Related searches
Another thing that can be obtained very easily from Google. If there are any questions with your key phrase, maybe it’s worth trying to answer them?
Database
Large information-gathering databases, such as Wikipedia or Wikidata, also suggest related articles. What’s more, we are able to find a lot of synonymous expressions for our phrase.
Knowledge graph
For some search results, Google displays a text box with the most important information and associations. It’s also a good source for potential references and related concepts for our keywords.
Summary
Search engines still lack a lot to understand exactly what the user might mean, but they are finding newer and newer ways to deal with it. Learning the mechanisms and algorithms that allow them to do this will certainly be useful to future content creators.
Leave a Reply