Using machine-learning model to generate a user interface with personalized combined filters for search results

A computer program product comprising a non-transitory computer readable storage medium having instructions encoded thereon that, when executed by a processor, cause the processor to perform steps comprising:

• obtaining, by a computer system, a collection of filters, each filter from the collection of filters having a respective functionality for filtering search results;
• receiving, at the computer system, a search query entered via a user interface (UI) of a device associated with a user of the computer system;
• generating, using the received search query, a set of search results;
• responsive to the received search query, obtaining a set of features for the user;
• generating, based at least in part on the set of features for the user, a set of candidate filter combinations, each candidate filter combination from the set of candidate filter combinations having combined functionalities of a plurality of filters from the collection of filters;
• accessing a filter scoring machine-learning model that is trained to predict, for each candidate filter combination from the set of candidate filter combinations, an engagement likelihood for the user or a conversion likelihood for the user, wherein the engagement likelihood for the user is a likelihood that, given selection by the user of a first UI element having a functionality of a first filter of the plurality of filters, the user will select a second UI element having a functionality of a second filter of the plurality of filters, and the conversion likelihood for the user is a likelihood that, given selection by the user of the first UI element and the second UI element, the user will convert on an item from the set of search results;
• applying the filter scoring machine-learning model to the set of features for the user, a set of features of the search query, and a set of features for the plurality of filters associated with each candidate filter combination to generate a score for each candidate filter combination from the set of candidate filter combinations that is indicative of the engagement likelihood for the user or the conversion likelihood for the user;
• selecting, using the score for each candidate filter combination from the set of candidate filter combinations, a set of filter combinations personalized for the user that are the most likely to be used by the user among the set of candidate filter combinations;
• generating, based at least in part on the selected set of filter combinations and the set of search results, a UI signal;
• and
• sending, via a network, the UI signal to the device associated with the user, wherein sending the UI signal causes the device associated with the user to display the UI with a set of UI elements personalized for the user along with the set of search results, each UI element from the set of UI elements having a functionality of a respective filter combination from the selected set of filter combinations for filtering the set of search results when each UI element is selected by the user, for generating a refined set of search results using the functionality of the respective filter combination, and for updating the UI to display the refined set of search results.

A computer system comprising:

• a processor;
• and
• a non-transitory computer-readable storage medium having instructions that, when executed by the processor, cause the computer system to perform steps comprising:
• obtaining, by the computer system, a collection of filters, each filter from the collection of filters having a respective functionality for filtering search results;
• receiving, at the computer system, a search query entered via a user interface (UI) of a device associated with a user of the computer system;
• generating, using the received search query, a set of search results;
• responsive to the received search query, obtaining a set of features for the user;
• generating, based at least in part on the set of features for the user, a set of candidate filter combinations, each candidate filter combination from the set of candidate filter combinations having combined functionalities of a plurality of filters from the collection of filters;
• accessing a filter scoring machine-learning model that is trained to predict, for each candidate filter combination from the set of candidate filter combinations, an engagement likelihood for the user or a conversion likelihood for the user, wherein the engagement likelihood for the user is a likelihood that, given selection by the user of a first UI element having a functionality of a first filter of the plurality of filters, the user will select a second UI element having a functionality of a second filter of the plurality of filters, and the conversion likelihood for the user is a likelihood that, given selection by the user of the first UI element and the second UI element, the user will convert on an item from the set of search results;
• applying the filter scoring machine-learning model to the set of features for the user, a set of features of the search query, and a set of features for the plurality of filters associated with each candidate filter combination to generate a score for each candidate filter combination from the set of candidate filter combinations that is indicative of the engagement likelihood for the user or the conversion likelihood for the user;
• selecting, using the score for each candidate filter combination from the set of candidate filter combinations, a set of filter combinations personalized for the user that are the most likely to be used by the user among the set of candidate filter combinations;
• generating, based at least in part on the selected set of filter combinations and the set of search results, a UI signal;
• and
• sending, via a network, the UI signal to the device associated with the user, wherein sending the UI signal causes the device associated with the user to display the UI with a set of UI elements personalized for the user along with the set of search results, each UI element from the set of UI elements having a functionality of a respective filter combination from the selected set of filter combinations for filtering the set of search results when each UI element is selected by the user, for generating a refined set of search results using the functionality of the respective filter combination, and for updating the UI to display the refined set of search results.