Since its beginning, Google has employed simple but effective algorithms to deliver results. PageRank, Google’s first algorithm, served up results based on backlinks. The higher the number of backlinks a website has, the more relevant or important that website is.
With RankBrain, it seems clear that Google’s future lies in artificial intelligence.
What is Google RankBrain?
In October 2015, Bloomberg broke the story that a machine-learning AI, nicknamed RankBrain, would contribute to serving up Google’s search results.
How does RankBrain work?
Whenever RankBrain sees a word or phrase it doesn’t recognize, it uses similar words and phrases with similar meanings and then filter out the irrelevant results. It does this by converting written language into mathematical language, or vectors, that the AI system can understand.
In short, RankBrain draws connections between words and phrases to interpret more complex or uncommon searches, without exact keyword matches.
While RankBrain won’t replace search engines anytime soon, it’s confirmed to be the third most important ranking factor, behind backlinks and content. This means that in addition to having strong content and weblinks that mention your webpage, RankBrain is the greatest influence in how you rank on SERP.
There are potentially thousands of different ranking factors, or “signals”, which can influence a result ranking, including PageRank, Google’s original algorithm. These signals all factor into Google’s main algorithm, called Google Hummingbird.
While it’s unclear exactly how many pages are ranked because of RankBrain, Greg Corrado, a senior research scientist at Google, says that it’s a “very large fraction” of the millions of queries searched each second.
How long has Google been working with AI?
This isn’t Google’s first rodeo with AI. In 2011, Google Brain, a deep learning research project, was developed. Their first software system was called DistBelief, a machine learning system used for Google search (voice and text), Photos, Youtube, Maps, and Street View.
The second-generation of DistBelief, called Tensor Flow, included a machine-learning library. Tensor Flow was made open-source in November 2015.
In 2014, Google bought out DeepMind Technologies, a British-based firm experimenting with artificial intelligence. DeepMind’s current research is focused on machine learning-computers capable of “learning for themselves directly from raw experience or data.”
As we store and collate more information than ever before, we’ll need to find new approaches to organize and present it. Gone are the days where we relied on teams of people to analyze user behavior. We’re witnessing the advent of machine-learning, systems capable of understanding complex information, such as the nuance of human language.
Arcalea combines marketing professionals and data analysis experts into a single team. We love reading the data, the challenge of "how can we?" and of continuously striving to raise our teammates and client partners to be the best they can be.