The meaning of the new algorithm Korolev
Posted: Wed Jan 22, 2025 8:19 am
Yandex Algorithm - Korolev
SEO
SEO
Evgeniy Letov
Yandex's "space" broadcast about the launch of the new "Korolev" algorithm has ended. What conclusions can we draw a minute after the end of the speeches?
"Korolev" is a Palekh search algorithm, rolled out not only on the page title, but also on the text of the document.
Yandex Algorithm - Korolev
Content
Video of the launch of "Queen"
Help on semantic algorithms
The meaning of the new algorithm Korolev
How to promote under the Korolev algorithm
What topics did "Korolev" touch on?
Market reaction to the announcement of the "Korolev" algorithm
Video of the launch of "Queen"
The video tells about what was done by Yandexoids before launching and including the Yandex Korolev algorithm in public search.
Help on semantic algorithms
Officially from Yandex about Korolev , about Palekh . Full description of processes for students of the Mathematics and Mechanics Department on Habr .
Yandex determines the semantic similarity of the texts of a document and a user's query based on billions of assessors' ratings. More than a million assessors (Tolokers) contribute to the development of search in Yandex Toloka . By the way, anyone can post their tasks for assessment in Toloka.
How it works in Russian. Pay attention to the query [ cats in space ] in Image Search. By the way, there have never been cats in space. And there are no texts that contain pictures for this query and text about it. However, the results for this query are normal. Yandex, based on assessors' ratings, has learned to understand the meanings of pictures == texts.
Example of the Korolev algorithm - cats in space
The assessors' ratings are analyzed by a neural shareholder phone number data find connections between the document texts (words, uni-, bi-, trigrams, etc.) and the user's request.
Now the document does not necessarily have to contain keywords. Yandex will understand what the text is about anyway. It will analyze how semantically close the words of the text are to the queries. And it will put this document higher than it was before the introduction of "Koroleva".
The so-called "query index" is also taken into account - the history (or rather semantic proximity) of search queries that users previously used to access the site. By the way, Dmitry Sevalnev spoke about the existence of a query index, no lie, in 2013 (There was a domain about fur coats. They replaced the content, started promoting furniture on it - and nothing fit at all. They changed the domain - furniture is at the top).
SEO
SEO
Evgeniy Letov
Yandex's "space" broadcast about the launch of the new "Korolev" algorithm has ended. What conclusions can we draw a minute after the end of the speeches?
"Korolev" is a Palekh search algorithm, rolled out not only on the page title, but also on the text of the document.
Yandex Algorithm - Korolev
Content
Video of the launch of "Queen"
Help on semantic algorithms
The meaning of the new algorithm Korolev
How to promote under the Korolev algorithm
What topics did "Korolev" touch on?
Market reaction to the announcement of the "Korolev" algorithm
Video of the launch of "Queen"
The video tells about what was done by Yandexoids before launching and including the Yandex Korolev algorithm in public search.
Help on semantic algorithms
Officially from Yandex about Korolev , about Palekh . Full description of processes for students of the Mathematics and Mechanics Department on Habr .
Yandex determines the semantic similarity of the texts of a document and a user's query based on billions of assessors' ratings. More than a million assessors (Tolokers) contribute to the development of search in Yandex Toloka . By the way, anyone can post their tasks for assessment in Toloka.
How it works in Russian. Pay attention to the query [ cats in space ] in Image Search. By the way, there have never been cats in space. And there are no texts that contain pictures for this query and text about it. However, the results for this query are normal. Yandex, based on assessors' ratings, has learned to understand the meanings of pictures == texts.
Example of the Korolev algorithm - cats in space
The assessors' ratings are analyzed by a neural shareholder phone number data find connections between the document texts (words, uni-, bi-, trigrams, etc.) and the user's request.
Now the document does not necessarily have to contain keywords. Yandex will understand what the text is about anyway. It will analyze how semantically close the words of the text are to the queries. And it will put this document higher than it was before the introduction of "Koroleva".
The so-called "query index" is also taken into account - the history (or rather semantic proximity) of search queries that users previously used to access the site. By the way, Dmitry Sevalnev spoke about the existence of a query index, no lie, in 2013 (There was a domain about fur coats. They replaced the content, started promoting furniture on it - and nothing fit at all. They changed the domain - furniture is at the top).