Google's BERT in 2 minutes

August 11, 2020
2 min read

BERT (Bidirectional Encoder Representations from Transformers) is a deep learning algorithm related to Natural Language Processing (NLP) that helps Google Search to understand what words in a sentence mean with all the nuances of context [1]. In English, BERT helps Google understand the intent behind a search query.

As an example, let’s say you search “I love chicken”. This doesn’t mean that you have romantic feelings towards chicken but you need to eat a chicken. With BERT, Google knows that you want to eat chicken and it will show KFC or McDonalds or shops that serves chicken, not shops where you buy gifts for your girlfriend! What BERT essentially does is, it reads the surrounding words of a certain word to understand the intended meaning of that word. To do so, BERT uses Google NLP services like Syntax Analysis, Sentiment Analysis, Entity Analysis, etc. and combine the results together to come to a conclusion.

Live example:

A user searched for “math practice book for adults”. Before BERT, Google couldn’t process the phrase “for adults”. So, it just showed up math practice books. Now with BERT, Google knows the person is looking for math books for grown-ups. [2]

I want BERT on my site. How?

You CANNOT! Google Search Liaison, Danny Sullivan tweeted, “There's nothing to optimize for with BERT”. But then the Webmaster Trends Analyst at Google, John Mueller offered an explanation with that can be done to SEO for BERT. [3]

Write the content naturally using a simple language.
Creating highly specific, Succinct content around a topic.
Be Specific with the Keywords or Search Queries you target.
Using Topic Clusters (Topic Cluster is a group of interlinked web pages that are constructed around a pillar content which targets a broad topic [4]).