Bloomberg news broke the story that Google was using a new system called RankBrain, to help with search engine queries.
RankBrain is a machine learning artificial intelligence that can help process searches within their set of algorithms. Specifically, it seems that the AI uses pattern recognition to help decipher the millions of searches that “confuse” the search engine.
While there are hundreds of algorithms that go into a Google search, RankBrain is impressive because it is designed to be a learning system that can help refine even the most complicated of searches. It is also the latest in Google’s ongoing effort to understand natural language on a higher level.
From a business perspective, it is unclear what RankBrain will do for SEO and search rankings. As of right now, the system is used to help parse language and does not have an effect on where results will appear on the page. But from a technical perspective, this has the potential to change everything when it comes to Entity Searches.
What is an Entity Search?
Entity search, to put it simply, is taking a person’s search query and relating it to an idea or object, as opposed to looking at it as a string of words. For instance, when you type in “Mazda”, Google will give you a list of possible search terms with the term “Mazda” in it as recommendations. This has made searching for things on Google much easier, but it still relies on being able to attach a set of search results to an idea, which can get confusing if the search engine does not know what kind of idea you are trying to look up.
For instance, if you throw a string of words together, such as “Car Champ Ski Goggles”, it will turn up all kinds of random results. That is because the Search Engine is trying to use what it knows as ideas to find what you are looking for. RankBrain’s ability to parse language and learn from it will help Google link it’s results with more ideas, even the more abstract ones. So as time passes, that string of words will make sense to Google. This will mean more accurate searches. At least, this is the hope.
by Bruce Worrall, Nov-10-2015