“Yandex” improved search using neural networks
[ad_1]
The Russian company “Yandex” updated the search engine using the CS YATI neural network, trained on documents for IT specialists and assessments of programming experts. The company claims that this made query navigation more convenient, and search results for developers and ML specialists better.
Thanks to the new search engine, programming-related queries now take into account one and a half times more information on the page than before. The CS YATI neural network was trained on terabytes of programming documents and learned to predict the clicks of skilled programmers in order to produce the most relevant response.
Also, for the convenience of developers, the Stack Overflow enriched answer system has been updated: now the question itself, the best answer to it, and other comments that may be useful to programmers will be displayed directly in the search results. The display of snippets (fragments of the source text under the link in the search engine) has also been improved, more useful information is added there.
About the directions of development of “Yandex” – in the material “Search bar – Latin America”.
[ad_2]
Source link
