BERT – Google’s Latest Update Can Change Online Search Results in a Big Way

The Content Team 4 years ago

Indeed, Google searches have become integral to our daily lives. There is hardly any day where you don’t use this world’s most popular search engine to find answers to our routine queries. In fact, internet search has become akin to phrase “Google it!” Techies at Google are also constantly striving to improve search engine algorithms so that the results are closer to the actual requirement of the user. BERT is a welcome move in this respect.

Concise Intro

Whenever you ask Google to search based on words, phrases or sentences, there is a specific intention behind it. The reason may however not always be apparent to the search engine algorithms currently in use. Google can overcome this issue. It can understand the user’s intention or derive the actual meaning behind a search string through the Bidirectional Encoder Representations from Transformers (BERT) algorithm. Hailed by Google as the most important update in last five years, here is more into it.

More about BERT?

To start with, BERT is an open-source algorithm used to understand the meaning behind a text search query. It can handle several tasks related to natural language processing such as:

  • Entity recognition
  • Question-answering, and
  • Parts of speech tagging

Due to the open-source nature, BERT has many versions and all of them intend at improving the results for user searches based on Natural Language Processing (NLP).

According to the official release by Google, these improvements aim at understanding the meaning or intention of the person using the Google Search function. This is where BERT chips in and helps Google to derive the nuances of conversational language queries and come up with better results.

For example, we know that the use of prepositions like ‘for’ and ‘to’, can alter the meaning of the whole sentence, their use is helpful in understanding their actual meaning during conversations. BERT is fulfilling a similar function for Google Search; it will help it by grasping the meaning behind conversational searches. Thus, it will enable us to use phrases or queries that we are comfortable with, instead of trying to mould our queries for easier understanding by Google Search and consequently, losing result accuracy.

How AI is constantly improving Google Search?

As you can now infer, Google is increasingly using machine-based learning to constantly improve the quality of results for its users. Thus, smart algorithms like BERT can now derive the meaning of entire phrases used in a particular context. This is a huge improvement over understanding individual words – as their combination can mean something entirely different. The complexity of English language further exacerbates the issue, making it difficult for machines to understand human speech and word patterns. BERT is a significant step in resolving the issue, even though it is still far from perfect. Currently, it can derive the intended meaning behind even the misspelled search phrases, much like us humans can understand jumbled sentences. As of now, BERT is able to affect about one in ten Google Searches, which is quite significant with billions of queries processed every day.

Wrapping up!

So, Google Search now functions even better with the help of BERT algorithm. The algo can understand conversational texts better and display pages that match the searched information more accurately. But, is this technology the precursor to self-aware, self-improving artificially intelligent machines? Only time will tell.

Stay in touch with Whitevox

Sign up for our monthly newsletter and follow us on social media for the latest news.