Technologies
Yandex Search and other products rely on many technologies.
In 2009 Yandex launched MatrixNet, a new method of machine learning. It allows Yandex's search engine to take a very large number of factors into account when ranking search results. MatrixNet also allows customization of ranking formulae to a specific class of search queries. For example, music searches may be fine-tuned without undermining the quality of ranking for other types of queries.
In July 2010 Yandex developed and implemented a real-time search. Yandex Search learned to recognize search queries that refer to the latest events and a new searchbot named Orange was launched for real-time indexing.
Yandex continued to expand throughout 2010 with a new search technology named Spectrum. Once again the idea behind it was to understand what the user may want. If a query is ambiguous, the system will use query statistics to guess the user's intents. So, if the majority of users searching for Gone with the Wind expect to find a film, the majority of search results will be about the film, not the book.
Yandex has also developed a method to search for and categorize duplicate images. The technology was implemented in Yandex's image search product, Yandex.Images. When it finds duplicated copies, it categorizes them into four groups: exact duplicates, thumbnail duplicates, semi-duplicates and enhanced semi-duplicates.
Another Yandex service, Yandex.Traffic, calculates the average levels of city traffic congestion using data from drivers who use the mobile version of the Yandex.Maps service. The technology processes GPS tracks and merges all available information on traffic jams.
Read more about this topic: Yandex