Search recommendations with neural networks

The algorithm is designed to display products that are most suitable for user's search query. Recommendations are based on neural networks that use the behavior of other users with similar search queries and also take into account the meaning of search phrases. Search recommendations with neural networks should be applied for the internal search results page and in case of absence of search results.

🚧

Important

The ability to use the algorithm depends on the amount of statistics collected on search queries and the total volume of offers available. To clarify information on the use of the algorithm on a specific store, please contact your personal manager.

📘

Link

link consists of a name and a value, in the query string the name and value are separated by a comma, example: link=card,12345, where card is the name and 12345 is the value.

🚧

Order of product IDs

The order in which product identifiers are issued in the algorithm is important: they are returned sorted in descending order of relevance according to the requested algorithm.
Therefore, they must be rendered in the same order.

Language
Click Try It! to start a request and see the response here!