Abstract: In one embodiment, a method includes receiving a query; generating a query embedding representing the query corresponding to a point in an n-dimensional embedding space; identifying one or more image objects matching the query; accessing, for each of the identified image objects, an image embedding representing the image object corresponding to a point in an m-dimensional embedding space; transforming, using a relevance model, the query embedding and each of the image embeddings into a joint p-dimensional embedding space; calculating, for each identified image object, a relevance-score based on a similarity metric between the transformed query embedding and the transformed image embedding; generating search results based on the calculated relevance-scores; and sending, to the client system in response to the query, instructions for presenting a search-results interface to the user, wherein the search-results interface includes search results referencing identified image objects presented in ranked order based

Abstract: In one embodiment, a method includes a computing system receiving a query from a first user, which can be parsed to identify i conditions associated with the query. The system may then identify one or more search results substantially matching the i conditions. Each search result may be associated with a feature vector of j features. The system may then access a conditioned ranking model that comprises j scoring functions for each i condition. The j scoring functions may correspond to j features of the feature vectors. A score for each search result may be calculated based on the i conditions and the j features. The system may then receive a selection of one of the search results from the first user, and in response modify one or more of the j scoring functions of the conditioned ranking model based on the selection.