BERT: Google's new algorithm affects the delivery of results in search engines

BERT: Google's new algorithm affects the delivery of results in search engines

BERT

One in 10 queries on Google search engines will be affected by the arrival of the new BERT algorithm.

This new feature, which will be fully operational in Search globally very soon, is the biggest change since the launch of the Rank Brain algorithm five years ago.

The big news is that BERT places increasing emphasis on natural language and this also further segments search by profile.

What is BERT?

The BERT (Bidirectional Encoder Representations from Transformers) algorithm was launched globally last week by Google and will soon be fully operational in all languages ​​around the world.

BERT will help search engines better understand the nuances and context of searches, and understand questions as a human would to provide more relevant answers.

The algorithm comes with a more specific function so that search engines can better understand the word, which can bring ambiguities and different meanings in the most varied languages.

The word alone may mean one thing, but as the sentence evolves, that meaning may change.

Rank Brain remains active

Although it provides a significant change by analyzing search queries and providing an additive understanding of search engines, BERT will not replace Rank Brain, which analyzes the content of pages and queries in the Google index, that is, web pages.

Google will decide which algorithm will provide the most relevant answers to the user. At some point it may use BERT, at others it may use Rank Brain.

The arrival of BERT also makes keyword optimization quite unlikely, because Google is giving too much importance to natural language. In other words, keyword relevance will decrease.

BERT

According to Dawn Anderson, who is one of the world's greatest SEO experts, in natural language the relevance is not for structured data, but for understanding what wants to be clarified in the query.

For her, the meaning of words has to do with their context. That's why BERT, especially in voice queries or audio marketing, which is another trend within digital marketing, will analyze tone of voice, feelings and many other aspects that encompass the understanding of natural language.

This means that, despite the questions being the same, the same results are not always delivered to different people, because the answers will be presented based on the analysis of a true set of factors.

How to have content for natural language?

But what content should be created for natural language analysis? For Dawson, the main tip is to create headings that promote word disambiguation, that is, avoid those that have confusing meanings. For example, the word “rose” can mean several things, so context becomes relevant with BERT.

Furthermore, it is also necessary to always think about better accessibility, which ensures that content can be accessed and understood by any user.

Therefore, using clear structures can help transform unstructured data into semi-structured data. Using lightweight content pages will also help improve search engine rankings.

To help you prepare even more appropriate content, a Google Cloud tool can also be very useful. Called Natural Language, this feature helps you check whether your content complies with natural language analytics in Search.

You need to pay attention to these changes that will be promoted in the delivery of Search results to understand if your website has been impacted by them. It is important to remember that natural language will be increasingly valued by artificial intelligence.

 

Leave the first comment

Contents