Google Rank Brain
From our previous post on CJG Digital Marketing, we talked about the latest and biggest announcement from Google and their reliance from machine learning and artificial intelligence to handle some of its search results. Now we know how it helps us to refine our sometimes-ambiguous search queries, let us now wrap up the top 8 most important facts surrounding this newest talk of the town- the RankBrain.
1 IT LEARNS OFFLINE
All of the RankBrain’s learning process takes place offline. According to Google, It uses three mechanisms as follows:
RankBrain establishes how to make predictions based on historical data it processed. Predictions are then tested and proven before they go can go online. The cycle of team-offline-and test is repeated each time a new search query is entered into the system.
2 IT IS SMARTER THAN GOOGLE’S SEARCH ENGINEERS
It is not doubtful that Google itself is the one of the home of the world’s finest and brightest software engineers. You’d think that all the time they spent creating and tweaking algorithms and updates since the day of Google’s inception would give them the ability to outsmart an al like RankBrain. But that is not the case. In one experiment conducted by Google to measure the effectiveness of RankBrain’s ability to show the most relevant content, the company put its search engineers head-to-head with RankBrain.
During the test, RankBrain achieved a score of ……80%
While Google’s search engineers scored lower at…..70%
One of the complex query use in the test and was given in the Bloomberg article was:
What’s the title of the consumer at the highest level of a food chain?
Now. to an amateur person the word ‘consumer’ sounds like a reference to someone who buys something, but it is also a scientific term for something that consumes food.
From the context of the question, it is clear that it pertains to the highest level in the food chain which is predator This are the sort of questions the RankBrain is apparently helping, according to Google.
3 IT’S NOT THE SAME AS PAGERANK
PageRank was one of the algorithms created by Google founders Larry Page and Sergey Brin, and was applied rly in the company’s history.
Since then The PageRank algorithm has always been one of the main components for Google’s Hummingbird and also serve as the main backbone lor Google search lor years.
Now that Google has unleashed its newest tool. The question of whether this machine marks the demise of PagesRank still remains.
Earlier this year. Google has decided to kill off Toolbar PagesRank but explained that the company still uses its data internally inside the ranking mechanism.
A blog by Jonathan Bright suggested that PagesRank slow demise is somehow a “good riddance” becuse SEO should always go with the quality and not quantity.
Rather than looking for good ranking factor, which is likely to be the case of PageRank, RankBrain base the result on the accuracy and the relevance of the content it is presented with.
4 IT HANDLES AROUND 15% OF GOOGLE’S DAILY SEARCH QUERIES
Google’s search engine is all powerful, but it does not hold all the knowledge in the world.
According to Google’s own estimate, on a daily basis, the search engine processes over….
3 BILLION Search queries around the world, and…..
15% submitted are queries Google have never seen before.
We have to solve the problem* the lead designer for Coogle Search John Wiley said in an Interview with Bloomberg TV. Is comprised of unique searches such as ambiguous queries, colloquial terms, and voice search and conversational speeches that RankBrain helps to better interpret and effectively translate.
5 IT CONVERTS WORDS INTO VECTORS.
Vectors are mathematical representation of how words are related with each other. The value of each word is achieved by feeding RankBrain plenty of words to analyze.
After this, if will then learn the semantic relationship between words and group of words.
There is also a blog about vectors posted by Coogle last 2013, and even talks about how a machine learned about the concept of capital cities and their corresponding countries just by reading lots of news articles and with no human supervision involve.
6 RANKBRAIN UTILIZES THE LOWEST FORM OF ARTIFICIAL INTELLIGENCE
According to an article by TechCrunch there are generally 3 classifications for Artificial Intelligence:
Artificial Narrow Intelligence
Artificial General Intelligence
Artificial Super Intelligence
When we are talking about RankBrain’s machine learning algorithm, we are actually pertaining to Artificial Narrow Intelligence.
It sounds surprising judging from the amount of searches it handles every day. But quality doesn’t necessarily equate to Intelligence since RankBrain has only one particular thing to do which Is to refine unique search queries.
7 MICROSOFT’S BING HAS SOMETHING SIMILAR.
Back in 2005. Microsoft released its own learning machine called RankNet, and it appears to function very much comparable to RankBrain
A blog by John Lincoln of Ignite Visibility even suggested that Google might probably got the Idea for RankBrainIrom Bing’s RankNet But after many years, the people in the tech industry has barely heard about RankNet.
It’s main proponent. Mr.Chris J.C Burges was recently honored for their 2005 paper that showed how RankNet works.
8 IT’S THE GOOGLE’S DETERRENT TO FACEBOOK’S SEARCH FYI
In the late 2015, Tom Stocky, Vice President of Search at Facebook, announced on its blog that it’s on its process of updating its Search FYI.
Stocky’s team made it possible to search all the public post- which currently stands at 2 trillion, on Facebook by developing a new function that uses algorithm that rank and refine trillions of posts by Facebook users.
Similar to RankBrain’s objective to find semantic relationship on words and alike, Facebook search tool will direct you to results- photos and posts from friends and strangers, that it thinks it most relevant to your query.
But according to an interview by Bloomberg to Mark Mahaney.an analyst at RBC Capital Markets, the odds are very slim for Facebook to successfully take over Google as the people’s primary choice lot searching online and “they have to come up with a better search experience because as of now. People think of Facebook as a place to communicate and share things with their friends and not as a search tool.