Google RankBrain – Now the third-most important ranking signal

Over the last few months, roughly 15% of Google searches have been interpreted by the new Google system known as RankBrain.

RankBrain uses a type of AI called Deep Learning which converts words and phrases into mathematical objects known as vectors. These vectors allow the Google computers to understand the words and phrases in terms of probabilities and patterns. This approach is particularly useful when the user has either miss-spelled a word or has used a word or phrase which the Google algorithms have not seen before.

In tests at the Google search laboratory, RankBrain was pitted against the Google search engineers to guess which from a group of pages would rank top for specific phrases. The Google engineers guessed correctly 70% of the time, while RankBrain achieved an impressive 80% success rate.

RankBrain is one of the roughly 300 signals which are used in the Google search algorithm.  In an interview with Bloomberg Greg Corrado said “in the few months it has been deployed, RankBrain has become the third-most important signal contributing to the result of a search query.  I was surprised, I would describe this as having gone better than we would have expected.”

The includsion of RankBrain in the algorithm is part of Google’s overall strategy to invest heavily in the AI side of the company in order to stay ahead of the competition.  The company is looking to expand AI and Deep Learning technology into every aspect of it’s business.  Google’s CEO Sundar Pichai added “machine learning is a core transformative way by which we are rethinking everything we are doing.”

In order to understand a little more about Deep Learning and how RankBrain learns and deals with ‘vectors’, read the Google post “Learning the meaning behind words.”  This post describes how a computer learns through similarities in concepts. The chart from the article depicts how after reading thousands of news articles, Deep Learning can allow a computer to learn the concept of capital cities.

country-capital-vectors

The team behind RankBrain:

greg-corradoGreg Corrado – A Google senior research scientist interested in biological neuroscience, artificial intelligence, and machine learning. His Google publications include “Building high-level features using large scale unsupervised learning” and “Large Scale Distributed Deep Networks.

 

 

yonghuiYonghui Wu – A specialist in natural language processing and machine learning, his work includes using AI methods in a variety of medical situations in order to build new software tools and to extract information from enormous data-sets including predicting adverse drug reactions, analysing hospital discharge summaries and ranking relationships between genes and drug treatments.

 

thomas-strohmannThomas Strohmann – A senior software engineer involved in the Google Knowledge Vault which moved away from text-based data extraction and into a probabilistic approach using large existing knowledge stores and machine learning.

 

 

What can the SEO industry learn from RankBrain?

The impact of RankBrain on the SEO industry is still a little unclear, but from the information Google has released so far, we can assume that rankings for sites which are aggressively targetting known keywords and phrases may lose out to websites where the prime objective is to create unique and interesting content.

We believe that as the technology behind AI and Deep Learning progresses websites should move towards being an authority on all aspects of the niche that they occupy, writing rich content that customers enjoy and placing less focus on keyword stuffed webpages.

 

Tags:
,
1Comment
  • Nikolay Stoyanov
    Posted at 13:49h, 21 April Reply

    Pretty nice article. I was intrigued by something. On this graph, there are distances between the countries and cities. How do you think that these distances are created? Is this based on how often two countries are mentioned in random, internet articles or perhaps it is due to political and economic cooperation?

Post A Comment