Google is a search engine that works very well. Born in 1998, it has been crawling and indexing the web ever since. Google that has more than 200 ranking factors to determine a web page relevant to a search query, is adding another one: RankBrain.
The company explained it as a machine learning Artificial Intelligence (AI) to help how its search engine works.
Google confirmed its existence on 26 October 2015, and in an interview, the company commented that it is the third most important factor in its ranking algorithm, coming after links and contents.
With the growing internet, countless of pages are introduced to the World Wide Web in any given day. RankBrain here uses AI to embed that vast amount of languages into mathematical entities which called vectors. These vectors help Google to see the correlation of a given query to another existing queries, which may have a meaning in relation.
What it does, is to learn whether or not a word or phrase is familiar, or known, to its algorithm. If not, Google can then guess words or phrases that might have similar meaning. Then the search engine can use its algorithms to filter the result accordingly.
AI has come to numerous fields in technology. As for Google Search, it is used to process search results in order to provide more relevant results by becoming more effective at handling search queries never heard by the search engine.
Google never disclose how its 200 different ranking factors work. The company keeps its secrets conceal in order to prevent people abusing it and taking advantage of its weaknesses. And here, RankBrain is just another way Google approaches the web with its arsenal and careful planning.
Related:
How Google Search Works, And How It Can Show You The Things You Want
How Search Engines Process Your Queries Determines Your Satisfaction
Keeping Up With The World
Google is good in doing its job. But how will AI be when aiding the already existing calculation depends on the progress of AI development itself. At one point-of-view, it's a relevant proof that Google is getting itself more dependent on machine learning to help it sort the web. In other words, Google is becoming a tech company that believes computers to soon replace humans and normal computation methods. Its ultimate goal in to develop its core search algorithm based on it. The biggest advancement with RankBrain is in how Google deals with the quantity of content it analyzes in order to create the vectors.
RankBrain is bigger than the classic "link anchor text and surrounding text" many SEO experts are always considering when discussing optimization.
Over the recent history, Google has a substantial percentage" of all queries handled by AI. And RankBrain is becoming one part of it. The goal is to make RankBrain able to create a search environment where the web becomes an easier and better place.
But how RankBrain will affect users?
At first, Google users weren't noticing anything radically different on how Google shows its search result.
By understanding the meaning of a query by relating it others, the more RankBrain considers a web document to have a potentially correct answer to an unknown or not understandable query. This in turn will put that document at a higher rank in the corresponding SERP - while of course still taking into account the other applicable ranking factors.
The company is planning to make its database of information the base of RankBrain's knowledge repository. This means that Google is allowing RankBrain to learn humans and their way of conducting search by observing the data flow.
RankBrain that uses mathematical processes and an advanced understanding of language semantics to gradually learn more about how and why people search, should be able to apply those conclusions to future search results.
So rather than being pre-programmed to respond to certain situation (like how Google usually was), it can update itself over time.
In short, Google is aiming RankBrain to limit human intervention. While it's still buggy as most AIs are, Google wants it to not rely on any input (pure machine learning). It will only need a dataset, over which it will apply its learning process in order to generate and refine its algorithm.
But on the other side, Google is also putting more flaws to the web.
How is that possible? AI is still at an early phase of development. It's still great on its own, but has a lot to catch up in terms of understanding human behavior and the way of interacting. While AI comes with many"forms", we have seen some great examples of good AI gone bad. One of which was Microsoft's Tay.
Making RankBrain To Think Like Humans But Works Like Machines
Keeping up with the trend is one thing, the next is to keep up with the number of information it has to give. As the internet is growing, machine learning has a huge advantage because it uses computer's capabilities to "learn"; it teaches itself to do something rather than having humans teaching it what to do.
Consider this: traditional way of computation is great to process data in a flow that is predictable. Tens, hundreds or even thousands of data at any given time could be processed flawlessly with it. By when considering million, billions or beyond, the flow is becoming unpredictable: Things won't be just a simple input - output anymore. Humans may have the ability to control the flow, but the limit on a computer's brain that is able to process data a lot quicker than humans, has a huge advantage in data processing.
As a part of Google's overall search algorithm, RankBrain is that computer program that's used to sort the billions of pages it knows, and to find those than deemed most relevant to particular search queries.
While Google has numerous algorithms in its disposal, such as the Hummingbird, RankBrain here is just another contender. All of its algorithms work together with PageRank that covers a specific way of giving pages credit based on the links from other pages pointing at them.
As a conclusion, people can consider RankBrain to be the first 100 percent post-Hummingbird algorithm developed by Google that helps it understand "verbose queries." Similar to the Hummingbird, it generalizes and rewrites those kinds of queries, trying to match the intent behind them.