What is semantic search?
It’s probably easier to give you examples rather than trying to paraphrase the rather verbose Wikipedia entry.
Remember how ‘back in the day’ you would type a search query into Google and it would hit you with a result based purely on your keywords?
Like if I had typed in ‘Portishead’ without context I would have been offered a mixture of results based on the band and possibly the town in Somerset.
However right now in this modern semantic world of ours, if I type in ‘Portishead’ I’m purely given results about the band, mainly because most of my search history is music related and not at all Somerset related.
That’s semantic search at its most simplistic. Taking various details about the user’s history and providing the most relevant results.
However semantic search can do much more than this. Semantic search can take multiple variables about the person typing in the search query and use these to serve more accurate and relevant results.
It’s all about moving from delivering results purely on the user’s intent to taking into account the entire context of the search.
By intent we mean a standard query such as ’steak restaurants’ and the context can mean anything from the location of the user, the time of day, their search history and what device they’re using.
Google will also take into account other people’s search queries from around the same time and the relative success of those results.
More examples of semantic search
My experience with semantic search over the last couple of days has been nothing less than impressive (or spooky if you want to look at it another way). All it takes is a trip abroad to reveal its true power.
I’m currently in New York. Yeah it’s no big deal, it’s much the same as London. We have just as many fistfights on public transport and people walking into you because they’re looking at their mobile phones as you do.
Before travelling I had to fill out a travel authorisation document (or ESTA) online.
After going through the whole process, I was told that I already had one outstanding from last year. My next instinct was to begin typing a question into Google ‘how long does an ESTA last?’ but I barely had to type more than the first two words before Google answered my question before I’d already asked it.
Switching to incognito mode reveals that when people normally type ‘how long’ they’ll be served results based on the most popular queries.
So in the context of my question, Google knew that I had spent considerable time on the ESTA website, and based on other people’s experiences and contexts (UK based, has visited the ESTA website before, has flown to the USA afterwards, has returned to the ESTA website within two years afterwards) provided me with the most likely search result. And it was entirely correct.
Later that same day...
Because I use Gmail, and this is where I receive all my flight information, Google served me with all my relevant (and up-to-date) flight information, even though I had merely searched for my departing airport Heathrow.
Alright Google, you can quit following me around now. It’s getting weird.
How does it work?
In 2013, Google launched the Hummingbird algorithm. This was the point when you may have noticed Google becoming way more precise and instinctive when answering your search queries.
As TechCrunch reported at the time, Hummingbird was the biggest overhaul to its search engine since 2009 and this brand new algorithm allowed Google to more quickly analyse full questions (as opposed to analysing searches on a word-by-word basis) and to identify and rank answers to those questions from the content Google has indexed.
Any single word is now understood to have multiple meanings by Google, and the way these words are contextually used by anyone who has ever used Google before, means that far more accurate results can be delivered. Even before a user has finished their query.
Google is using all of this search data to build its own knowledge base, you may have heard of this referred to as the Knowledge Graph.
The Knowledge Graph is a system for organizing and merging information about millions of well-known people, places, and organizations from many data sources, including Wikidata.
When this was introduced in 2012, the Knowledge Graph resulted in simple information cards like the contemporary one below, featuring links and other info pulled in from Wikipedia.
Now however, the Knowledge Graph has become even more sophisticated and prevalent. Here’s my search for ‘best albums of 2015’ which results in this giant carousel across the top of the page.
This leads me to my next question…
Why does Google care so much about building such a vast resource of information?
It’s in its own mission statement…
Google’s mission is to organize the world’s information and make it universally accessible and useful.
The non skeptical among you will understand the desire to make the internet a better, more useful, user-focused place. It’s what keeps me writing about this stuff and it’s at the heart of most successful digital transformation endeavours.
Of course the rest of us can argue that the better Google can understand its users, and the quicker it can deliver relevant accurate content, the longer users will spend on Google. Therefore increasing the chance that a user will click on an ad.
You can help Google find and deliver your content by marking it up as accurately as possible using semantic HTML, just bear in mind that some of these techniques (star ratings, images, having your own search box appear in your search results) may make your results more attractive, but are also designed to keep the user on Google for as long as possible.
The more questions Google can answer itself, using your own content, the less users will actually interact with you site.
Well that ended way more downbeat than I intended. I'll cheer myself up by asking Google where's the nearest store I can buy a wiener dog.
To view the original article Click Here