It’s difficult to remember where search technology was only 10 years ago. After all, 2010 really doesn’t feel that distant, does it?
But things have changed. Dramatically.
10 years ago everyone was focussed on backlinks and keywords.
Today, you should be focused on search intent, and user behaviour. And this is largely due to how far semantic search has come since it first rocked our world back in 2012.
What is Semantic Search?
Semantic search, as the name suggests, is all about meaning.
It relates to a search engine - Google or otherwise - attempting to decipher the meaning behind any given search based on intent, context, and the relationship between words.
Search engines have, in recent years, become immeasurably better at understanding what it is you are searching for and serving as much relevant content as possible, as quickly as possible.
It’s all about user experience. Isn’t everything?
Think back 10 years. The onus was for more upon you, the searcher, to think carefully about what you were searching to get the answer you were looking for as quickly as possible.
An Example of Semantic Search
According to IMDB, Dwayne ‘The Rock’ Johnson has 79 acting credits to his name. And I think we all know there’s an awful lot of garbage in there.
Not wanting to waste your time, you search “what is Dwayne Johnson’s best film?”
Before 2013, when the Hummingbird update was released (we’ll get onto that in a bit), Google would have simply sought to match the keywords in the searched phrase and would have given you webpages featuring those exact keywords.
In 2021 we get so much more...
In 2021 we get a list of, presumably, Dwayne’s most popular movies, more information about the man himself (a knowledge panel) as well as a link to his Wikipedia page, two suggested follow-up questions, as well as a link to his best films as ranked by Rotten Tomatoes.
What more could you want?
More than that, the sophistication of Google means that it is able to interpret searcher intent based on a range of factors, including:
User search history
Global search history
What are the Principles of Semantic Search?
As we said, semantic search is all about meaning. Consequently, there are two primary principles that guide semantic search.
The first is search intent. Why is someone performing that search? Common types of search intent are informational, navigational, and transactional - i.e. learn, find or buy.
Being able to distinguish between those three types of intent means that search engines are able to provide the most relevant results to the searcher - i.e. an answer, the location of a shop, the best price for the latest Nike Vaporfly...
The second is semantic meaning. According to Google, semantics relates to “the branch of linguistics and logic concerned with meaning.”
Extrapolating that out to search, semantics relates to how a search queries, connected words and phrases, and the content of web pages crossover.
By considering the connections that exist between these three areas, search engines are able to serve results that speak to the context of the words used to search, and not only their literal meaning.
Why is Semantic Search Important?
Naturally, this boils down to user experience. Again, doesn’t everything?
Semantic search is important for a number of reasons.
Firstly, people speak and search differently, both locally and globally. The way two people on the same street use Google may differ drastically, let alone someone in the US versus a searcher living in Thailand or India.
Language, background, personal experience, tone, sentence construction. There’s so much diversity to overcome, and semantic search helps us to overcome it.
And this connects to the ambiguity of search, our second point. Understanding the context in which someone is searching removes some - or, ideally, all - of that ambiguity.
Finally, the words and entities (a thing or concept) included in a sentence are all related to one another. The meaning of one word can change drastically based on the words around it.
Semantic search is important for understanding those relationships.
The Origins of Semantic Search
Semantic search is a topic in vogue right now. But it’s nothing new. In fact, it’s been around for nearly 10 years.
The Google Knowledge Graph (2012)
The first steps toward what we now know as semantic search started with the Google Knowledge Graph. Introduced in 2012, the Knowledge Graph amassed over 500 billion facts about five billion entities. It changed everything.
The introduction of the Knowledge Graph meant that Google was able to understand the relationship between information and billions of entities.
The Knowledge Panel was the first and is the most visible trace of the Knowledge Graph.
Google Hummingbird (2013)
Google’s big algorithm update of 2013 took things a step further.
With the introduction of Hummingbird, Google was not only able to understand the relationship between information and entities, it began to understand conversational search better.
This meant that rather than tailoring our searches for an algorithm, we could start searching in a more natural way, and Google would understand.
As Moz puts it, “Hummingbird signaled Google’s commitment to an increasingly sophisticated understanding of the intent of searchers’ queries with the goal of matching them to more relevant results.”
If it wasn’t already, the introduction of RankBrain is when Google’s algorithm became truly clever. Before RankBrain, Google’s basic algorithm did all the hard work matching queries to the most relevant results. And it did a pretty good job.
But RankBrain brought a sophisticated machine learning component to Google’s core algorithm that is far better at sifting through the billions (and billions) of data points on the web, and connecting the searcher with the most relevant information.
Again, as Moz puts it, “it is believed that the query now goes through an interpretation model that can apply possible factors like the location of the searcher, personalization, and the words of the query to determine the searcher’s true intent.”
And, of course, RankBrain is able to learn as a result of the information it is fed. Information is rarely static, and so the ability of the algorithm to learn has been hugely beneficial, even if you haven’t noticed it.
As a result of RankBrain, it’s easier than ever to find out what The Rock’s latest movie is.
And that brings us almost up to date, except for 2019’s BERT update.
Bidirectional Encoder Representations from Transformers (BERT) is the most important algorithm update, at least from a semantic search perspective, since RankBrain.
We’ve covered BERT in-depth previously. But, in essence, BERT made it possible for Google to better understand language. According to Google…
“BERT models can… consider the full context of a word by looking at the words that come before and after it—particularly useful for understanding the intent behind search queries.”
This post is intended as something of an introduction to semantic search, and there’s so much more that can (and will) be said on the topic. Not least because it’s a field that will continue to develop as the years wear on.
What’s clear is that search has come on leaps and bounds in the last 10 years, but almost (almost) without us noticing it. It’s difficult to remember a world where people needed lessons on how to search Google effectively, but they existed not so long ago.
Some of that is down to the inexorable progress of society. But much of it is down to semantic search, and the ability of Google (and other search engines) to understand a human, and not algorithmic, use of language.