Abstract: A system including a processor and non-transitory computer-readable media storing computing instructions that, when executed on the processor, perform a method that includes training a machine learning model by using a training data set to determine taxonomy embeddings for taxonomies based on training features that include a context word vector, a center word vector, and a probability. The probability corresponds to a function between the context word vector and the center word vector. The taxonomy embeddings represent at least a first level of a taxonomy and a second level of the taxonomy. The machine learning model, as trained, is used to determine the taxonomy embeddings based on taxonomy identifiers and reduce the taxonomy embeddings by removing at least one taxonomy of the taxonomies that are below a threshold to thereby reduce the taxonomies. The threshold comprises a number of aggregate page views.

Abstract: Systems and methods for automatically retrieving and providing digital advertisements from multiple channels with improved relevancy to a search query are disclosed. In some embodiments, based on a query, a first set of sponsored items is retrieved and ranked based on an optimization of conversion rate, and a second set of sponsored items is retrieved and ranked based on an optimization of click-through rate. Based on the first set of sponsored items and the second set of sponsored items, a ranked list of recommended items is generated for display based on an advertisement auction mechanism, in response to the query.

Abstract: A system including one or more processors and one or more non-transitory computer readable media storing computing instructions that, when executed on the one or more processors, perform: receiving a taxonomy identifier corresponding to a taxonomy for a product; determining taxonomy embeddings based on the taxonomy identifier, the taxonomy embeddings representing at least a first level of the taxonomy and a second level of the taxonomy; modifying taxonomies based on a threshold to reduce a number of the taxonomy embeddings in subsequent processing; and mapping the taxonomies, as modified, to publisher placements to display the product within the taxonomies on a graphical user interface (GUI).

Abstract: Systems and methods for automatically retrieving and providing digital advertisements from multiple channels with improved relevancy to a search query are disclosed. In some embodiments, based on a query, a first set of sponsored items is retrieved and ranked based on an optimization of conversion rate, and a second set of sponsored items is retrieved and ranked based on an optimization of click-through rate. Based on the first set of sponsored items and the second set of sponsored items, a ranked list of recommended items is generated for display based on an advertisement auction mechanism, in response to the query.

Abstract: Systems and methods of generating interfaces including variant item recommendations are disclosed. A request for an interface and a set of candidate items selected from an item catalog are received. At least one of the candidate items is representative of two or more variant items. A variant score is determined for each variant item related to the at least one of the candidate items and a set of recommended items is generated by independently ranking each item in the set of candidate items and each of the two or more variant items. The set of recommended items is generated by a variant-aware ranking model configured to receive the variant score for each variant item and based on the variant score and a relevancy score for each variant item. The interface including the set of recommended items is generated and transmitted to a system that generated the request for the interface.

Abstract: Systems and methods for identifying an audience set are disclosed. A request receive identifying a future time period and item class is received and a conversion value for each of a set of user identifiers is generated by implementing a trained statistical machine learning model using historical transaction data. A first subset of user identifiers and a second subset of user identifiers are identified based on threshold values of the conversion value. The subsets are each associated with targeted advertisement types corresponding to a particular level of specificity associated with the requested item class. The first targeted advertisement type is presented to user devices associated with the first subset of user identifiers and the second targeted advertisement type is presented to user devices associated with the second subset of user identifiers.

Abstract: A system including one or more processors and one or more non-transitory computer readable media storing computing instructions that, when executed on the one or more processors, perform: receiving a taxonomy identifier corresponding to a taxonomy for a product; determining taxonomy embeddings based on the taxonomy identifier, the taxonomy embeddings representing at least a first level of the taxonomy and a second level of the taxonomy; modifying taxonomies based on a threshold to reduce a number of the taxonomy embeddings in subsequent processing; and mapping the taxonomies, as modified, to publisher placements to display the product within the taxonomies on a graphical user interface (GUI).

Abstract: A method implemented via execution of computing instructions configured to run at one or more processors and stored at one or more non-transitory computer-readable media. The method can include receiving, via a computer network, an ad request. The method also can include retrieving ad candidates from an ad database. The method further can include determining a respective ad ranking score for each of the ad candidates, based at least in part on the ad request and respective historical retrieval scores for each of the ad candidates.

Abstract: Systems and methods for selecting items from a pool of items based on comparisons of composite embeddings are disclosed. An anchor embedding is generated for a target string. The anchor embedding is a composite embedding including at least a first initial embedding and a second initial embedding. An item embedding is obtained for each item in a pool of items. Each item embedding is a composite embedding including a first initial item embedding and a second initial item embedding. A similarity score is generated by comparing the item embedding to the anchor embedding for each item in the pool of items and a set of items is selected from the pool of items. The set of items includes a predetermined number of items in the pool of items having a highest similarity score.

Abstract: Systems and methods for selecting items from a pool of items based on comparisons of composite embeddings are disclosed. A first initial embedding is obtained from a first database for each item in a pool of items and a second initial embedding is obtained from a second database for each item in the pool of items. The first initial embedding is generated using a first embedding model and the second initial embedding is generated using a second embedding model. A first composite embedding is generated for each item in the pool of items comprising the first initial embedding and the second initial embedding; and compare the first composite embedding for each item in the pool of items to a first anchor embedding, wherein the first anchor embedding comprises a first initial anchor embedding generated using the first embedding model and a second initial anchor embedding generated using the second embedding model.

Abstract: This application relates to apparatus and methods for determining digital product advertisements for products a customer is more likely to purchase. Historical order data is obtained for orders previously placed by a customer. In some examples, a first value is determined for a first brand based on a purchase date of any of the orders for the customer that include at least one item of the first brand. A second value is determined based on the first value and the purchase date of any orders that include at least one item of a second brand. A brand affinity score for the first brand is determined based on the first value and the second value. In some examples, the brand affinity score is based on a machine learning process. The brand affinity score may determine what brand of a product the customer is more likely to purchase.

Abstract: This application relates to apparatus and methods for determining digital product advertisements for products a customer is more likely to purchase. Historical order data is obtained for orders previously placed by a customer. In some examples, a first value is determined for a first brand based on a purchase date of any of the orders for the customer that include at least one item of the first brand. A second value is determined based on the first value and the purchase date of any orders that include at least one item of a second brand. A brand affinity score for the first brand is determined based on the first value and the second value. In some examples, the brand affinity score is based on a machine learning process. The brand affinity score may determine what brand of a product the customer is more likely to purchase.

Abstract: This application relates generally to automated systems and methods to identify an audience set for a marketing campaign period. In an embodiment, a system includes at least one processor operatively coupled with a datastore, the at least one processor configured to receive, from a user device, a request identifying a time period and an item class. The at least one processor is further configured to retrieve, from the datastore, user identifiers based on the time period and the item class. The at least one processor is further configured to determine a conversion value for each of the user identifiers by applying a statistical model to historical transaction data associated with the user identifiers. The at least one processor is further configured to determine an audience set comprising a subset of the user identifiers with the conversion value exceeding a threshold value.