Abstract: In some implementations, a device may receive from a user device, a request. The device may determine, based on the request, a retrieval set that includes a plurality of candidate results for the request. The device may generate a plurality of request-result pairs based on the request and the retrieval set. The device may determine relevance outputs for the plurality of request-result pairs using a secondary model that is trained based on a primary model using: (1) an unlabeled dataset that is based on a candidate result set for a historical request, and (2) a margin-based loss function on all candidate result pairs of the candidate result set for the historical request. The device may transmit, to the user device and in accordance with the relevance outputs, data for presentation in a user interface of the user device based on the request.

Abstract: A system including a processor and a non-transitory computer-readable media storing computing instructions that, when executed on the processor, cause the processor to perform certain operations: splitting a set of candidate loads into subsets of candidate loads, wherein time windows of the subsets of candidate loads align with time ranges of a facility; iteratively reducing, without integer constraints, the subsets of candidate loads into a number of candidate loads; determining multiple candidate loads from the number of candidate loads that optimize an objective function with the integer constraints; and selecting at least one candidate load from the multiple candidate loads corresponding to a mode of transportation. Other embodiments are described.

Abstract: A method can include: storing, on an electronic platform, a vectorized item database corresponding to a plurality of items; tracking electronic interactions between the electronic platform and a user computer during a user session; generating, using an embedding model, a session context embedding based, at least in part, on the electronic interactions tracked during the user session; in response to receiving the search query from the user computer, querying the vectorized item database to identify search results corresponding to one or more of the plurality of items, at least in part, using the session context embedding; and transmitting, by the electronic platform, the search results to the user computer. Other embodiments are disclosed.

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, cause the one or more processors to perform operations: generating, using at least two different search algorithms, combined search results based on a search query, each search result of the combined search results having a respective rank based on at least one of the at least two different searching algorithms; determining a set of search results from the combined search results, wherein each result within the set of search results was respectively interacted with by at least one user; and determining a traffic impact for each result rank in the set of search results. Other embodiments are described herein.

Abstract: Examples relate to a computer-implemented method or system including a processor that can perform certain operations. The operations can include obtaining inputs comprising dockout windows for routes from a depot, capacity information for the depot, and dockout preferences. The operations also can include determining respective assignment penalties associated with respective assignments for feasible combinations of the routes and time periods within the dockout windows for the routes. The operations additionally can include determining a solution set of the respective assignments that minimize a total penalty. The operations further can include reassigning at least a portion of dockout times in the solution set to spread out the portion of the dockout times at the depot based on the dockout preferences. The operations additionally can include outputting the dockout times, as reassigned, for the routes. Other embodiments are described.

Abstract: Examples provide a system for filtering the contents of adjacent item storage compartments from item recognition results obtained using computer vision (CV) for more accurate mapping of item locations within a retail environment. A filter manager selects a vertical member and tracks it throughout a series of images generated by an image capture device. The selected vertical member defines at least a portion of a target compartment in each image. The vertical member can be a display case door or a vertical steel bar defining the side of an item storage compartment. A set of adjacent items located on each side of the target compartment is filtered from the item recognition results. The target compartment location is determined based on a location tag of the target compartment. The items remaining after filtering are mapped to the target compartment location while reducing CV item location false positives.

Abstract: Some embodiments provide a product inventory information distribution apparatus, comprising: a central electronic data storage buffer; a control circuit; and a memory storing computer instructions that when executed cause the control circuit to: detect product inventory update information messages, wherein each of the product inventory update information messages is received from one of multiple different product inventory systems and is generated by a corresponding one of the multiple product inventory systems when one of the multiple product inventory systems detects a change in an inventory state corresponding to any one of hundreds or more products available through the product inventory storage facility; direct each of the multiple product inventory update information messages to the central electronic data storage buffer; and maintain messages in the central electronic data storage buffer received during a buffer retention window of time.

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, cause the one or more processors to perform operations: parsing a hypertext markup language source code from text strings of query strings of a webpage containing ingredients for a recipe; extracting, using an iterative process, the ingredients for the recipe from the query strings of the webpage; mapping recipe products based on quantities or units of measure of the ingredients, as extracted, for the recipe; and automatically adding the recipe products to an electronic shopping cart. Other descriptions are disclosed.

Abstract: A device may receive a plurality of retrieval sets, first sampling parameters, and second sampling parameters that are based on offline sampling that uses multiple-metric efficacy values that are based on one or more efficacy metrics and one or more policy metrics. The plurality of retrieval sets may be from a plurality of clusters of a set of object embeddings representative of a plurality of objects. The device may select a retrieval set of the plurality of retrieval sets by performing online sampling using the first sampling parameters. The device may determine an object ranking in the retrieval set by performing online sampling using the second sampling parameters. The device may generate code or markup based on the object ranking in the retrieval set.

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, cause the one or more processors to execute operations: determining, using a weight determination model of a machine learning architecture, (a) token weights for tokens included in a search query and (b) extraneous tokens from among the tokens based on at least the token weights; comparing the token weights for the tokens to a probability threshold; segregating, based on at least the token weights, the extraneous tokens from a remainder of the tokens; after segregating, identifying first tokens of the remainder of the tokens that satisfy the probability threshold; and identifying, based on the first tokens of the remainder of the tokens that satisfy the probability threshold, search results for the search query. Other embodiments are described.

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, cause the one or more processors to perform operations: generating a preliminary dock-out time boundary based on one or more delivery time windows for one or more deliveries; processing two parallel data flow paths to locate a range of dock-out times associated with one or more time constraints of the one or more delivery time windows; and transmitting the range of dock-out times to a user, wherein a trailer is dispatched to a distribution center between the range of dock-out times to load the trailer and deliver the one or more deliveries within the time constraints of the one or more delivery time windows based on a flexible period of time. Other embodiments are described.

Abstract: This disclosure relates to computer vision and generative artificial intelligence (AI) techniques for generating a target image set based on content included in a source image. The target image set comprises a plurality of target images, each of which is compliant with or more target display specifications for an electronic platform. The target images can be generated by an image adaptation network that comprises various AI models, including one or more saliency model, one or more generative models, one or more scene detection models, and/or one or more segmentation models.

Abstract: According to an embodiment of the present disclosure, a system is provided. The system includes a processor and a non-transitory computer-readable medium storing computing instructions. The instructions, when executed on the processor, perform: receiving user search queries and product items, the product items including metadata corresponding to product types, the product types comprising an error, and the error corresponding to one of the product items having a non-compliant product type; modifying the metadata corresponding to the product types based on a product type group to mitigate a computing system from displaying non-compliant material to a user; and in response to a product type of the product type group being the non-compliant product type, performing a computer vision prediction on an image of a product corresponding to the product type to detect a misclassification in the product type group. Other embodiments are disclosed.

Abstract: Some embodiments include systems and methods for recall estimation. An exemplary method comprises determining reservoir sampling and size of labeling from each strata; determining recall and variance for each strata; determining sum of sample size for a time period; and determining reservoir sampling for each strata and label the sample items. Other embodiments are described.

Abstract: A system including a processor and a non-transitory computer-readable media storing computing instructions that, when executed on the processor, cause the processor to perform certain operations: identifying a first transit time of a first load that is different from a second transit time of a carrier for the first load; tuning a load generation model using a slackness allowance function; generating, using the load generation model with the slackness allowance function engaged, multiple slackness time windows with corresponding slackness penalties for the first load; and determining a third transit time based on a slackness time window of the multiple slackness time windows and a corresponding slackness penalty for the first load. Other embodiments are disclosed.

Abstract: This application is directed to systems and methods for detecting phishing attempts in a user application. In some embodiments, a disclosed method includes extracting from an incoming message a uniform resource identifier (URI) for identifying a resource on a computer network, generating a screenshot image of the resource identified by the URI, applying a phishing detection model to process the screenshot image and generate a phishing indicator representing a confidence level of determining that the resource would cause a phishing attack, and in accordance with a determination that the phishing indicator satisfies an alert condition, reporting via an alert message that the URI extracted from the incoming message corresponds to the phishing attack. In some embodiments, the alert condition includes a confidence threshold, and requires that the alert message be generated and reported in accordance with a determination that the phishing indicator is greater than the confidence threshold.