Google makes news display smarter with search–using “BERT”

«In the mobile version of" Google Search ", when a user in the US searches for a specific news topic, a carousel of articles related to the topic is displayed, and other related articles can also be confirmed.


Provided by Google

«Articles related to" Top Stories "[Top News] are displayed on the card under each carousel main card.According to GoogleThis feature allows users to determine which article to click after getting more detailed information.

Google explains in a blog post: “With the new structure, each topic is not only the most up-to-date article, but also high-quality content, and more space is allocated to articles from various sources. It becomes easy to spread

-Google recently improved the ability of "Google search" to understand what the person who is entering the query is looking for by using a model generated by the new machine learning technology "BERT"clearlydid.

BERT is an abbreviation for “Bidirectional Encoder Representations from Transformers”. It is used in the United States to process approximately 10% of English search queries.

The same BERT model is used to implement the newly developed “technology to understand articles” that Google has developed to “map people, places, and things related to news articles and draw out their relationships”. ing.

The purpose of such mapping is to create a group of news articles so that the time it takes for a user to get an answer after asking Google for a search service can be reduced. Starting December 11th, users using English in the United States will be able to use this feature on their mobile devices. It will be rolled out in more languages ​​and regions in the coming months.

This article is from overseas CBS InteractivearticleEdited by Asahi Interactive for Japan.

Source link

Show More

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button