Yandex has improved the search based on the CS YATI neural network


The Yandex search was renewed by using the CS YATI neural network. This model is trained on documents for IT professionals as well as evaluations by programming experts. As a result, the search output for ML specialists and programmers has become better, and the navigation on the request has become more convenient.

The new model is compared with its previous version – YATI, which took into account one and a half times less information. The updated neural network analyzed a huge number of output results and sites from programming-related queries. Due to the analysis of such an array of data, it better assesses the quality and relevance of the document. The neural network also learned to predict the clicks of qualified programmers to issue appropriate answers.

The press service of Yandex also reports on the refinement of enriched Stack Overflow answers. Now the user will be able to see the question, the best answer to it and other useful comments without going to the site. Also improved the display of snippets for GitHub and NPM by adding useful information there.

Yandex has improved search using the CS YATI neural network

Oleksiy Husakov, head of the machine intelligence and research department, notes that CS YATI studied mainly from English-language sources, since most requests from programmers are in English. Accordingly, the search for English-language sources was also improved.

Related Posts

Leave a Reply

Your email address will not be published.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

© 2022 - Theme by WPEnjoy · Powered by WordPress