Google has updated its search algorithm with technology BERT… Thanks to the innovation, the search engine, according to the developers, will better understand natural queries. An algorithm based on neural network technology must recognize the essence of a query, taking into account the peculiarities of each language, which will allow it to find the most relevant results for such queries.
In its blog, Google announces that its goal is to understand what a person wants to find on the global network and, regardless of the formulation of the request, to pick up the most relevant information on this issue. Thanks to the team s advances in language understanding through the use of machine learning technology, the search engine has made a key and largest leap in development in its entire history.
BERT Algorithm will be involved in ranking long search queries in “spoken” language using prepositions.
As an example of how the updated search engine will work, Google cited the query: “2019 brazil traveler to usa need a visa.” Until now, the search engine ignored the preposition “to”. This contributed to the fact that the search results featured information on how US citizens get into Brazil. Thanks to innovations that take into account language nuances, the search engine gives out relevant results.
It is planned that BERT will work in conjunction with RankBrain and affect 10% of requests. The technology is currently being used to recognize queries in English. In the future, the company will launch BERT for other languages.
Source: Google Blog
Follow us