Google revamps core search algorithm with natural language processing upgrade

0
665

Searching the Internet is practically synonymous with Google these days. As such, the Big Tech firm is constantly tweaking its search algorithm to help connect users with the information they need. Even between today and just five years ago, there is a massive difference in how Google returns search results.

Now, the company is ready to advance how the world searches one step further. Google is rolling out a huge upgrade to its core search algorithm that gives it the power to understand natural language processing. The new capability will help users find the results they are looking for faster and more efficiently than ever before.

Alphabet Soup

The world of search algorithms is highly complicated. Considering that Google processes 3.5 billion searches per day and 1.2 trillion searches annually, it’s safe to say that things are working well. Even so, the new upgrade will help make searching even easier.

Google engineers have been working on the natural language processing (NLP) feature for the past ten months and are now ready to roll it out across the platform. Essentially, it works by allowing the algorithm to understand how words in a sentence relate to each other.

Google’s previous search algorithm looked at a user’s query as a “bag of words.”

Now, it looks for context clues, word relationships, and more. At a recent press event, the company used the example query “Can you get medicine for someone pharmacy?”

The new search algorithm understands the context of these words. Rather than returning results for separate terms like “pharmacy” and “medicine,” it identified the phrase “for someone” and returned the appropriate results.

Meet BERT

Google’s new algorithm is based on Bidirectional Encoder Representations and Transformers (BERT). The system was trained by trying to figure out the missing 15 percent of words that Google removed from random English sentences. The Mad-Libs style learning model ultimately trained the system so well that it can now understand context.

Notably, Google claims that the upgrade to its algorithm will affect ten percent of all searches. Although the company says it doesn’t believe that the change will significantly alter where traffic flows, there could be some unintended side effects.

Of course, in a day and age where many companies earn a hefty portion of their revenue from search traffic, any decline related to the new algorithm could be harmful. As the decisions of machine learning algorithms are notoriously difficult to interpret, this could be an issue down the road. Nonetheless, Google wouldn’t be rolling out the update if it was even slightly concerned that it will cause problems.

Meanwhile, users should start seeing more accurate search results for their naturally-worded queries—even if they don’t think twice about it.