An online system selects an item to present to a user of the online system. The online system accesses user interaction data for the user. The online system transmits the user interaction data to a model serving system and receives, from the model serving system, item embeddings for the items with which the user interacted. The model serving system may use an LLM to generate the item embeddings based on the user interaction data. The online system generates a user embedding array based on the item embeddings. The online system applies a transformer network to the user embedding array to generate a user embedding describing the user. To select an item to present to the user, the online system compares the generated user embedding to item embeddings for a set of candidate items. The online system selects a candidate item based on the interaction scores.
An online system performs a message transformation task in conjunction with the model serving system or the interface system to transform a message input to a chat message. The online system receives the message input in a conversation between a picker and a customer. The online system may transform the message input to a text string that is properly formed and contextually appropriate, format the text string into a chat message, and send the chat message to a receiving party on behalf of the sending party.
G06F 40/58 - Utilisation de traduction automatisée, p.ex. pour recherches multilingues, pour fournir aux dispositifs clients une traduction effectuée par le serveur ou pour la traduction en temps réel
G06F 40/106 - Affichage de la mise en page des documents; Prévisualisation
G06F 40/166 - Traitement de texte Édition, p.ex. insertion ou suppression
G06F 40/232 - Correction orthographique, p.ex. vérificateurs d’orthographe ou insertion des voyelles
3.
RELATING ENVIRONMENTAL EFFECTS TO USER INTERACTIONS USING AUTOMATED SHOPPING CARTS
An automated checkout system applies environmental effects to physical regions within a store. The automated checkout system logs the environmental effect, a time the environmental effect was applied, and the physical region to which the environmental effect was applied. The automated checkout system detects an interaction event and logs a time associated with the interaction event. The automated checkout system identifies a location of the automated shopping cart and identifies a physical region within the store that contains the automated shopping cart's location. The automated checkout system identifies the environmental effect that was applied to the physical region at the time of the interaction event and generates a data point. For each environmental effect, the automated checkout system computes a success metric based on the generated data points. The automated checkout system applies environmental effects to physical regions based on the success metrics.
A system or a method for fulfilling orders using a machine-learned model in an online system. When a user places an order, the system accesses a model trained on historical data, including characteristics of candidate locations, previous orders, and recent inventory records. The model predicts the probability that each candidate location will incompletely fulfill the order. The system selects the location with the lowest probability of incomplete fulfillment and sends fulfillment instructions to client devices of available shoppers. After the order is fulfilled, the system receives data from the client devices of shoppers, identifies whether the order was completely fulfilled, and updates the machine-learned model based on the actual outcomes.
G06Q 10/0875 - Gestion d’inventaires ou de stocks, p.ex. exécution des commandes, approvisionnement ou régularisation par rapport aux commandes Énumération ou classification des pièces, des fournitures ou des services, p.ex. nomenclatures
G06Q 20/40 - Autorisation, p.ex. identification du payeur ou du bénéficiaire, vérification des références du client ou du magasin; Examen et approbation des payeurs, p.ex. contrôle des lignes de crédit ou des listes négatives
The present disclosure is directed to determining purchase suggestions for an online shopping concierge platform. In particular, the methods and systems of the present disclosure may receive, from a computing device associated with a customer of an online shopping concierge platform, data indicating one or more interactions of the customer with the online shopping concierge platform; determine, based at least in part on one or more machine learning (ML) models and the data indicating the interaction(s), a likelihood that the customer will purchase a particular item if presented, at a specific time, with a suggestion to purchase the particular item; and generate and communicate data describing a graphical user interface (GUI) comprising at least a portion of a listing of one or more purchase suggestions including the suggestion to purchase the particular item.
An online concierge system includes a content selection simulation module that performs offline simulations of a content selection process to enable rapid testing of various content selection parameters. The content selection simulation module obtains historical content selection data including content delivery opportunities and candidate content items associated with those content delivery opportunities. The content selection simulation module simulates the filtering, ranking, and auction stages of a content selection process using a set of configurable content selection parameters that affects selection of a winning content item and price. The winning content items from the simulation may be used to compute performance metrics associated with the configured content selection parameters. Different content selection parameters may be compared to determine an effect of changes to the parameters.
An online system provides a platform for users to place orders at different physical retailers. When a user moves from one location to another (e.g., the user physically moves or is traveling), where the user's preferred retailer is not available, the online system suggests a new retailer for the user and optionally items to purchase at the new retailer. When a user accesses the online system from a new location, the system obtains the user's previous purchases and computes a repurchase probability. The system then ranks candidate new retailers in the new location based on their match to the likely repurchased items. To suggest new items to buy at the new retailer, the system uses existing replacement models to suggest replacements for the items that the user is likely to buy based on previous purchases.
Based on logged information about prior events, an online concierge system generates a set of location metrics that quantify properties of locations such as retailers at which items may be acquired, and residences to which the items are brought. The location metrics can be used for a variety of purposes to aid customers or other users of the online concierge system, such as providing the users with more information (e.g., likely delivery delays) or alternative options (e.g., pricing options), or emphasizing options that the location metrics indicate would be of particular value to the user. To determine whether to emphasize a particular option, the online concierge system applies a machine-learned model that predicts whether emphasizing that option would effect a positive change in user behavior, relative to not emphasizing it.
Embodiments are related to automatic prediction of times for completion of tasks for an order by a picker associated with an online system and determination of an appropriate intervention for the picker. The online system applies a computer model to predict a plurality of times for completion of a plurality of tasks associated with the first order. The online system determines that the picker who accepted the first order did not complete a task of the plurality of tasks at a predicted time of the plurality of times increased by a threshold time. The online system determines an intervention associated with the picker, based in part on the determination that the picker did not complete the task. The online system causes a device of the picker to display a message that corresponds to the determined intervention.
An online concierge system uses a findability machine-learning model to predict the findability of items within a physical area. The findability model is a machine-learning model that is trained to compute findability scores, which are scores that represent the ease or difficulty of finding items within a physical area. The findability model computes findability scores for items based on an item map describing the locations of items within a physical area. The findability model is trained based on data describing pickers that collect items to service orders for the online concierge system. The online concierge system aggregates this information across a set of pickers to generate training examples to train the findability model. These training examples include item data for an item, an item map data describing an item map for the physical area, and a label that indicates a findability score for that item/item map pair.
A scrollable listing of icons associated with catalogs is displayed, in which the scrollable listing of icons is overlaid onto a page and remains fixed when the page is scrolled, each icon is associated with a catalog, and each icon is displayed with an indication of a set of items selected from a corresponding catalog. In response to a user selection of an icon from the scrollable listing of icons, the page is updated to include items included in a catalog associated with the selected icon. In response to a user selection to add an item from the page including the items, the indication displayed with the selected icon in the scrollable listing of icons is updated to indicate the added item.
Embodiments are related to using a trained computer model to predict a supply state of an online system for state-aware management of order fulfillments. The online system measures first values of a metric for a set of sample orders. The online system accesses the computer model trained to predict a value of the metric for an order placed with the online system. The online system applies the computer model to predict second values of the metric for the set of sample orders, based on one or more features of each sample order. The online system compares a distribution of the first values to a distribution of the second values and determines the supply state of the online system based on the comparison. Responsive to the determination of the supply state, the online system triggers a remedial action for the online system that adjusts the supply state of the online system.
An online system receives a request to confirm a transaction that is associated with an order. The system accepts or declines the transaction based on whether an amount associated with the pending transaction is likely to exceed an expected amount of the order by more than a threshold value. To determine the threshold, the system trains a first model to predict an overspend for an order and then trains a second model to predict an amount of error associated with the predictions from the first model. The outputs of the first model and the second model provide a mean and a variance for an expected distribution of the overspend. If the actual overspend amount for the transaction exists in too high of a percentile of the distribution, the transaction may be flagged for review or declined.
G06Q 20/40 - Autorisation, p.ex. identification du payeur ou du bénéficiaire, vérification des références du client ou du magasin; Examen et approbation des payeurs, p.ex. contrôle des lignes de crédit ou des listes négatives
14.
VALIDATING CODE OWNERSHIP OF SOFTWARE COMPONENTS IN A SOFTWARE DEVELOPMENT SYSTEM
A system validates code ownership of software components identified in a build process. The system receives a pull request identifying a set of software components. The system analyzes code ownership of each software component using machine learning. The system provides features describing the software components as input to a machine learning model. The system determines based on the output of the machine learning model, whether the code ownership of the software component can be determined accurately. If the system determines that a software component identified by the pull request cannot be determined with high accuracy, the system may block the pull request or send a message indicating that the code ownership of a software component cannot be determined accurately.
Embodiments relate to using a large language model (LLM) to generate a list of items at an online system with a user defined constraint. The online system receives a query that includes at least one constraint. The online system generates a prompt for input into the LLM, based at least in part on the query. The online system requests the LLM to generate, based on the prompt, a set of constraints for a set of item types. The online system generates a list of candidate items by searching through a set of items stored in one or more non-transitory computer-readable media using the set of constraints for the set of item types. The online system causes a device of the user to display a user interface with the list of items for inclusion into a cart, the list of items obtained from the list of candidate items.
Automatic creation of lists of items at an online system organized around co-occurrences of items. The online system provides inputs into a computer model, the inputs including information about items purchased by a user of the online system over a defined time period, information about a catalog of items stored at one or more computer-readable media of the online system, and a plurality of recipes each including a set of co-occurring items. The online system applies the computer model to generate an indication of co-occurrence of each pair of items in each recipe. The online system generates one or more lists of items based on the indication of co-occurrence, each of the one or more lists of items associated with a respective recipe. The online system causes a device of the user to display a user interface with the one or more lists of items for presentation to the user.
Embodiments relate to an automatic detection of fraudulent behavior for a transaction at an online system. The online system requests a large language model (LLM) to determine, based on a prompt input into the LLM, information about a refund event for a first order placed by a user of the online system. The online system accesses a computer model trained to detect a fraudulent behavior associated with an order placed with the online system. The online system applies the computer model to determine a score associated with the refund event, based on the information about the refund event received from the LLM. The online system determines, based on the score, whether the refund event was due to a fraudulent behavior of the user. The online system performs at least one action associated with the online system, based on the determination whether the refund event was due to the fraudulent behavior.
G06N 3/084 - Rétropropagation, p.ex. suivant l’algorithme du gradient
G06Q 10/087 - Gestion d’inventaires ou de stocks, p.ex. exécution des commandes, approvisionnement ou régularisation par rapport aux commandes
G06Q 20/40 - Autorisation, p.ex. identification du payeur ou du bénéficiaire, vérification des références du client ou du magasin; Examen et approbation des payeurs, p.ex. contrôle des lignes de crédit ou des listes négatives
Embodiments relate to determining an availability of a service option for delivery of an order placed with an online system. The online system receives an order placed with the online system. The online system accesses a computer model trained to predict a value of metric for an order placed with the online system. The online system applies the computer model to predict the value of the metric for the order. The online system determines which service option of a plurality of service options of the online system is available for delivery of the order, based at least in part on the predicted value of the metric and a threshold. The online system causes the device of the user to display an availability of the determined service option for delivery of the order.
G06Q 10/04 - Prévision ou optimisation spécialement adaptées à des fins administratives ou de gestion, p. ex. programmation linéaire ou "problème d’optimisation des stocks"
An online concierge system receives information describing items in orders placed by a customer and a sequence of events associated with each order and identifies an impulse item included in the orders based on a set of rules, attributes of each item, and/or the sequence of events. The system applies a model to predict a measure of similarity between the impulse item and each of multiple candidate items and identifies larger-size variants of the impulse item based on this prediction and attributes of the impulse item and each candidate item. The system applies another model to predict a likelihood the customer will order each variant, computes a recommendation score for each variant based on this prediction, and determines whether to recommend each variant based on the score. Based on the determination, the system generates and sends a recommendation for a variant to a client device associated with the customer.
Embodiments relate to order specific expansion of an area that encompasses pickers available for accepting an order placed with an online system. The online system accesses a computer model trained to predict an attractiveness metric for the order and applies the computer model to predict a value of the attractiveness metric for a first order. The online system classifies the first order into a first set or a second set, based on the value of the attractiveness metric and a threshold. Based on the classification, the online system expands over time a size of an area that encompasses a set of pickers available for accepting the first order. The online system causes a device of each picker in the set of available pickers located within the area of the expanded size to display an availability of the first order for acceptance by each picker in the set of available pickers.
Embodiments relate to automatic determination of a directed spend program eligibility for items offered by retailers associated with an online system. The online system provides inputs into a computer model, where the inputs include information about at least one property for each candidate item in a set of candidate items and at least one requirement for a directed spend program. The online system applies the computer model to generate, based on the inputs, an output that comprises an indication of an eligibility for each candidate item in the set for the at least one directed spend program. The online system sends a message causing a device of a user of the online system to display a user interface including an option for the user to add into a cart at least one of the candidate items determined to be eligible for the directed spend program.
An online system, such as a concierge service, provides services to users using a set of limited resources. To allocate the limited resources of the system among the users, the system uses a model to predict each user's sensitivity to different levels of service. An allocation module then allocates the limited resources among a set of users based in part on the estimated sensitivities and the supply of available resources.
An online system predicts replacement items for presentation to a user using a machine-learning model. The online system receives interaction data describing a user's interaction with the online system. In particular, the interaction data describes an initial item that the user added to their item list. The online system identifies a set of candidate items that could be presented to the user as potential replacements for the initially-added item. The online system applies a replacement prediction model to each of these candidate items to generate a replacement score for the candidate items. The online system selects a proposed replacement item and transmits that item to the user's client device for display to the user. If the user selects the proposed replacement item, the online concierge system replaces the initial item with the proposed replacement item in the user's item list.
An online concierge system generates a set of candidate estimated times of arrival (ETAs) for delivery of a set of items being purchased by a user. Each candidate ETA is scored by using a machine-learned model to estimate values for different criteria of interest, such as likelihood of acceptance of the ETA, cost of delivery of the items to the user, and the like. The values for the different criteria may be combined to generate the overall score for a candidate ETA. One or more of the highest-scoring ETAs are selected and provided to the user, who may then approve one of the ETAs for use with delivery of the user's set of items.
An online concierge system allows customers to place orders to be fulfilled by pickers. An order includes an amount of compensation a customer provides to a picker when the order is fulfilled. A customer may modify the amount of compensation provided to a picker, so some customers may initially specify a large amount of compensation to entice a picker to fulfill an order and then reduce the amount of compensation when the order is fulfilled. To prevent penalizing pickers who fulfilled an order without a problem, the online concierge system trains a model to determine a probability that a reduction in compensation to a picker was unrelated to a problem with order fulfillment. The online concierge system may perform one or more remedial actions for a picker based on the probability determined by the model.
G06Q 10/0875 - Gestion d’inventaires ou de stocks, p.ex. exécution des commandes, approvisionnement ou régularisation par rapport aux commandes Énumération ou classification des pièces, des fournitures ou des services, p.ex. nomenclatures
G06N 3/084 - Rétropropagation, p.ex. suivant l’algorithme du gradient
G06Q 20/12 - Architectures de paiement spécialement adaptées aux systèmes de commerce électronique
26.
EXTRACTING ITEM ATTRIBUTES FROM ITEM DESCRIPTIONS USING LARGE LANGUAGE MACHINE-LEARNED MODELS
A system may obtain an item description associated with an item in an item catalog. The system may generate a prompt for input to a machine-learned language model, the prompt specifying at least the item description and a request to identify one or more attributes of the item. The system may provide the prompt to a model serving system for execution by the machine-learned language model. The system may receive from the machine-learned language model, an output including a list of attributes and respective values associated with the item based on the item description. The system may standardize the formatting of the list of attributes and may store the list of attributes and the respective values for the list of attributes in association with the item in the item catalog.
An online system generates digital flyers using a generative model. The online system receives, from a client device, a request to generate a digital flyer. The request includes one or more design conditions for the digital flyer. For example, the design conditions may specify one or more cornerstone items, a theme, a template flyer, other target characteristics, etc. The online system further accesses an item catalog storing item data. The online system generates a query for a generative model including a prompt to generate the digital flyer, the one or more design conditions, and item data accessed from the item catalog. The online system provides the query to a model serving system, which executes the generative model with the query to return a batch of one or more digital flyers. The online system provides a first digital flyer in the batch of one or more digital flyers for presentation.
An online system predicts a number of interruption events within a time period and identifies anomalous numbers of interruption events using an interruption prediction model. The online concierge system maintains application state data that describes a state of an application workflow for a client application. The online concierge system identifies interruption events that represent interruptions to the application workflow and logs interruption events in an interruption log, wherein each entry of the interruption log describes an interruption event and its corresponding state. The online concierge system predicts a number of interruption events that will occur within a time period based on an interruption prediction model. The online concierge system computes an actual number of interruption events that occurred during the time period and computes a difference between the actual number and the predicted number. If the difference exceeds a threshold value, the online concierge system performs a remedial action.
Embodiments relate to automatically generating a basket of items to be recommended to a user of an online system. The online system communicates a basket opportunity to a group of retailers, wherein the basket opportunity defines a plurality of item categories each associated with a respective item to be included in a basket. The online system receives, from each retailer in response to the basket opportunity, a respective bid of a plurality of bids for the basket opportunity. The online system applies a computer model to each bid to determine a score for each bid and selects a winning bid for the user based on determined scores for the bids. For each item category, the online system populates the basket with a respective item from a catalog of a retailer that is associated with the winning bid. The online system then presents the basket with items to the user.
An online concierge system uses a machine-learned parking quality model to quantify the suitability of a particular parking location (e.g., a parking lot, or a street) for use when performing purchases at a retail location on behalf of customers. The parking quality model's output is determined according to input features related to parking at a candidate parking location, such as a current time, a current degree of demand for shoppers at the retail location, or a current average shopper wait time at the retail location before receiving an order. The online concierge system provides suggested alternate parking locations to a client device of the shopper, where they may be displayed, e.g., as part of an electronic map. Use of the suggested alternate parking locations helps to preserve parking availability in restricted areas such as retailer parking lots and to reduce traffic congestion in the area of the retailer.
An online concierge system performs asynchronous automated correction handling of incorrectly sorted items using point-of-sale data. The online concierge system receives orders from customer client devices and determines a batched order based on the received orders. The online concierge system sends the batched order to a shopper client device for fulfillment. The online concierge system receives transaction data associated with the batched order from a third party system. The online concierge system determines whether a sorting error occurred based on the transaction data and the batched order. In response to determining that a sorting error occurred, the online concierge system sends an instruction to correct the sorting error to the shopper client device.
G06Q 10/087 - Gestion d’inventaires ou de stocks, p.ex. exécution des commandes, approvisionnement ou régularisation par rapport aux commandes
G06Q 20/40 - Autorisation, p.ex. identification du payeur ou du bénéficiaire, vérification des références du client ou du magasin; Examen et approbation des payeurs, p.ex. contrôle des lignes de crédit ou des listes négatives
32.
User Interface for Obtaining Picker Intent Signals for Training Machine Learning Models
A concierge system sends batches of orders to pickers that they can review and accept in a batch list on a client device. Each batch in the batch list is presented with a hide option that enables the picker to hide a batch that they do not intend to accept. In response to receiving a hide signal, the system extracts features associated with the batch and stores those features with a negative indication of the picker towards the batch. The hide signal provides the system with a higher quality signal indicating the picker's negative intent regarding an order, as compared to simply ignoring the order in favor of fulfilling another order. This higher quality signal is then used to train models to better predict events related to the pickers' acceptance of orders, such as for ranking orders for pickers or for predicting fulfillment times.
An online system receives an indication that a user is starting an order. The online system retrieves candidate contents for the user and provides prompts to a model serving system. The model serving system is configured to provide scores for the contents based on relevancy, a likelihood of user interaction, and a likelihood of the user purchasing an item associated with the content. The online system provides scores from the model serving system to a predicted click-through rate (pCTR) model. Based on the pCTR model scores, the online system ranks the candidate contents. The online system provides content for display to the user based on the ranked candidate contents.
An online system determines whether to recommend a replacement item to a user based on a predicted sentiment score. The online system receives one or more comments from user feedback on the replacement items. The online system generates a prompt for each user comment for input to a machine-learned model. The online system generates a sentiment score for the ordered item and a replacement item based on the inferred sentiments by the model serving system. Using the sentiment score, the online system determines whether to recommend the replacement item.
An online system receives, from an entity, a content item to be presented to online system users, in which the content item includes a landing page to a third-party website. The system accesses the landing page, identifies a set of items included in it, and determines whether the landing page is configured for performing one or more types of conversions associated with each item. The system matches one or more of the items with one or more target objects based on the determination and associates the matched target object(s) with the content item. The system receives information describing one or more impression events associated with presenting the content item to a user and information describing a conversion associated with a target object associated with the content item performed by the user, applies an attribution model to determine a contribution of the impression event(s) to the conversion, and reports the contribution.
A computer system receives an image from a picker, which indicates an out-of-stock target item and potential replacements items. The system provides, to a machine learning model, a prompt requesting identification of the target item and the potential replacement items in the image. The system receives identification of the target item and a list of potential replacement items in the image. The system generates a first list of potential replacements items based on the list of potential replacement items identified in the image and a second list of replacement items from the target item by applying one or more replacement models to the identified target item. The system may merge the two lists and assign replacement scores to each item in the merged list to create a list of recommended replacement items. The system generates a message based on the image and the list of recommended replacement items.
G06V 10/774 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p.ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]; Séparation aveugle de source méthodes de Bootstrap, p.ex. "bagging” ou “boosting”
37.
Display panel of a programmed computer system with a graphical user interface
An online system uses a machine learning based language model, for example, a large language model (LLM) to identify replacement items for an item that may not be available at a store. The online system receives a request for an item and determines that the requested item is not available. The online system identifies a replacement item. If the online system determines that the replacement item has a replacement score below a threshold value indicating a low quality of replacement for the requested item, it uses a machine learning based language model, for example, a large language model to generate an explanation for why the replacement item has a replacement score below the threshold value. The online system sends the explanation to a client device.
INCREMENTALLY UPDATING EMBEDDINGS FOR USE IN A MACHINE LEARNING MODEL BY ACCOUNTING FOR EFFECTS OF THE UPDATED EMBEDDINGS ON THE MACHINE LEARNING MODEL
An online concierge system uses a model to predict a user's interaction with an item, based on a user embedding for the user and an item embedding for the item. For the model to account for more recent interactions by users with items without retraining the model, the online concierge system generates updated item embeddings and updated user embeddings that account for the recent interactions by users with items. The online concierge system compares performance of the model using the updated item embeddings and the updated user embeddings relative to performance of the model using the existing item embeddings and user embeddings. If the performance of the model decreases, the online concierge system adjusts the updated user embeddings and the updated item embeddings based on the change in performance of the model. The adjusted updated user embeddings and adjusted updated item embeddings are stored for use by the model.
A system maintains a data store for managing machine-learning (ML) models and features that are used by the models. The system generates a graph including nodes for each model and a node for each feature, and edges linking models and features that are used by the models. For a new model to be trained, the system receives a proposed feature corresponding to a node in the graph, and identifies one or more candidate features corresponding to nodes in the graph based in part on relevancy scores between the proposed feature with other features corresponding to nodes in the graph. The system presents in a user interface a suggestion to use one or more candidate features with the new model. Responsive to receiving a user selection of at least one candidate feature, the system causes the new model to be trained using the selected candidate feature and the proposed feature.
A feature management system (the “system”) receives information about a new machine learning (ML) model to be trained. The information includes metadata about the new model. The system applies a trained feature prediction model to the information about the new model and metadata about a plurality of features. The feature prediction model is trained to predict a probability that each of the plurality of features should be selected as an input feature for the new model. The feature management system identifies one or more candidate features based on an output probability score of the feature prediction model. The system presents in a user interface a suggestion to use the one or more candidate features with the new model. The system selects at least one candidate feature and causes the new model to be trained using a set of input features, including the selected candidate feature.
An inventory interaction model predicts user interactions with items of a location for a physical warehouse included with other warehouses in a region. The location is described with features that include the nearby locations and the respective user interactions with the respective item assortments, so that the item interactions for the evaluated location incorporate location-location effects in model predictions. To effectively train the model in the absence of prior interaction data for a location, training examples are generated from existing locations and user interaction data of item assortments by selecting a portion of the locations for the training examples and including nearby location interaction data, labeling the training example output with item interactions for the location. The trained model is then applied for an item assortment at a location by describing nearby locations in evaluating candidate locations and item assortments.
An online system evaluates different item assortments for a physical warehouse having limited capacity to stock items. Each item assortment is stocked at the physical warehouse in proportion to an assortment split weight. The items at the warehouse are available for users to order, for example to be gathered by a picker and physically delivered to users near the warehouse. Rather than display all items actually stocked at the physical warehouse to all users, the different item assortments are displayed to different users. Users may order items from the assigned item assortment and, because both item assortments are actually stocked at the physical warehouse, orders from either item assortment may be successfully fulfilled for delivery. The different user interfaces thus permit evaluation of the preferred item assortment by users while maintaining expected delivery capability and while using the same storage capacity of the physical warehouse.
For each retailer location associated with multiple retailers, an online system associated with the retailers receives video data captured within the retailer location by a camera of a client device associated with an online system user. The online system detects, based at least in part on the video data, a location associated with the user within the retailer location and/or an interaction by the user with an item included among an inventory of the retailer location. The online system generates a set of signals associated with the user based at least in part on the detection of the location and/or the interaction. Based at least in part on the set of signals, the online system determines a set of preferences associated with the user, trains a machine learning model to predict a metric associated with the user, and/or sends content for display to a client device associated with the user.
An online concierge system allows users to place orders for fulfillment by pickers. Orders have various attributes (e.g., dimensions, weight, contents, etc.), and the pickers may have corresponding characteristics affecting capability of fulfilling orders. To optimize allocation of orders to pickers for fulfillment, the online concierge system trains an order validation model that predicts a probability of a picker encountering a problem fulfilling an order based on characteristics of the picker and attributes of the order. The order validation model is trained from training examples based on previous orders and labels indicating whether a picker encountered a problem with fulfilling the order. The order validation model can then be used to predict deliverability of future orders or to specify limits on one or more attributes of orders for fulfillment.
An inventory interaction model predicts user interactions with items to be included in an item assortment in a warehouse. The item is described with features that include the co-located items and the respective user interactions, so that the item interactions for the evaluated item incorporate item-item effects in its predictions. To train the model effectively in the absence of prior interaction data for an item, training examples are generated from existing item and user interaction data of co-located items by selecting a portion of the items for the examples and including co-located item data, labeling the training example output with item interactions for the item. The trained model is then applied for an item assortment by describing co-located item features of the item assortment in evaluating candidate items.
In an online concierge system, a shopper retrieves items specified in an order by a customer from a retail location. The online concierge system optimizes order fulfillment by selecting a retail location for an order that is most time-efficient and that is most likely to have each of the item in the order available. Hence, the online concierge system may select a less convenient retail location that is more likely to have each item being ordered available. To predict whether a retail location incompletely fulfill the order if selected to fulfill the order, the online concierge system trains a machine learning model based on prior orders fulfilled by the retail location, a shopper retrieving items in the order, items in the order, and other features.
G06Q 10/0875 - Gestion d’inventaires ou de stocks, p.ex. exécution des commandes, approvisionnement ou régularisation par rapport aux commandes Énumération ou classification des pièces, des fournitures ou des services, p.ex. nomenclatures
G06Q 20/40 - Autorisation, p.ex. identification du payeur ou du bénéficiaire, vérification des références du client ou du magasin; Examen et approbation des payeurs, p.ex. contrôle des lignes de crédit ou des listes négatives
SELECTING A WAREHOUSE LOCATION FOR DISPLAYING AN INVENTORY OF ITEMS TO A USER OF AN ONLINE CONCIERGE SYSTEM BASED ON PREDICTED AVAILABILITIES OF ITEMS AT THE WAREHOUSE OVER TIME
An online concierge system allows users to order items from a warehouse, which may have multiple warehouse locations. The online concierge system provides a user interface to users for ordering the items, with the user interface providing an indication of whether an item is predicted to be available at the warehouse at different times. To predict availability of an item model at different times, the online concierge system selects data from historical information about availability of items at one or more warehouses based on temporal, geospatial, and socioeconomic information about observations of historical availability of items at warehouses. The online concierge system accounts for distances between observations and a time and geographic location in a feature space to select observations for predicting item availability at the time and the geographic location.
An online system receives information identifying items associated with a brand, a hierarchical taxonomy of the items, and information identifying a retailer associated with the brand. The system applies a machine learning model to predict availabilities of the items at (a) retailer location(s) associated with the retailer, identifies items that are likely available at the retailer location(s), and groups the identified items into categories based on the taxonomy. The system computes an item score for each item based on its popularity, attributes, and/or attributes of a user. The system assigns items in each category to positions within a display unit associated with the category and computes a category score for each category based on the item scores. The system assigns display units associated with the categories to positions within a template based on the category score and generates a page associated with the brand and retailer based on the assignments.
Selecting an Attribute of an Item for DIsplay in an Interface Based on Information Gain Determined for the Attribute by a Trained Machine-Learned Model
An online concierge system presents items to a user in one or more interfaces and maintains various attributes for each item. To optimize information about items in an interface, when the online concierge system receives a request for an interface, the online concierge system determines a context for the interface and a set of items to display in the interface from the request. For an item displayed by the interface, the online concierge system applies a trained attribute selection to each combination of the item, an attribute of the item, and the context for the interface to determine an information gain to the user from displaying the attribute of the item along with the item in the interface with the context. Based on the information gains, the online concierge system selects an attribute to display in the interface in conjunction with the item.
A search module for an online concierge system executes searches in response to a search query with respect to item databases of retailers. The search module dynamically configures a recall set size that controls a number of search results returned for a search query based in part on a query entropy representing an estimated breadth of the search term. The query entropy may be determined relative to a diversity of items in a retailer's database. The recall set size may be configured relative to the query entropy in a manner that manages a tradeoff between latency of search execution and search result quality.
An online system generates session-based recommendations for a user accessing an application of the online system. The online system receives, from one or more client devices, a sequence of actions performed by a user during a session of an application of an online system. The online system generates a sequence of tokens from the sequence of actions by tokenizing an action to a token representing a respective item identifier. The online system applies a transformer-based machine-learned model to the sequence of tokens to generate predictions for a set of items. The online system selects a subset of items based on the generated predictions for the set of items. The online system generates one or more recommendations to the user from the selected subset of items and displays the recommendations to the user.
An online system receives a request from a user to access an ordering interface for a retailer and identifies a retailer location based on the user's location. The system uses a machine learning model to predict availabilities of items at the retailer location and identifies anchor items the user previously ordered from the retailer that are likely available. The system computes a first score for each anchor item based on an expected value associated with it and/or a likelihood the user will re-order it, determines categories associated with the anchor items, and ranks the categories based on the first score. For each category, the system identifies associated candidate items likely to be available and ranks them based on a second score for each candidate item computed based on a probability of user satisfaction with it as an anchor item replacement. The ordering interface is then generated based on the rankings.
An online system performs an inference task in conjunction with the model serving system or the interface system to continuously monitor conversations between users and shoppers to determine whether a message sent by a sending party can be automatically responded to rather than prompting the receiving party for a manual response. The online system automatically provides a response to the message without the receiving party's manual involvement. In one or more embodiments, the online system can infer whether a question can be intercepted and/or suggests one or more available answers the sender can consider as feedback without a manual response from the receiver.
An online system employs real-time and pre-generated images for recommendation. The system leverages generative machine-learning models, such as diffusion models, to generate images dynamically. The selection and creation of these images rely upon user data and session data, which are collected during a user's application session. These data are employed to generate a text prompt string, which directs the image generation process. For instances where real-time computation may be a resource constraint, the system utilizes pre-generated images linked to user-context clusters—data set groupings related to user characteristics and session context. This method enables the system to present tailored recommendations to the user, making use of both dynamic generation and pre-existing image resources, thereby optimizing the balance between customization, computational resources, and latency.
An online concierge system receives an order from a user including items to obtain from a retailer for delivery to a location. A picker selects the order and obtains items from the retailer. The user selects a time interval during which items from the order are delivered to the location. To prevent the user from selecting a time interval for fulfillment the online concierge system prevents the user from selecting a time interval when a picker may be unable to obtain the items from the retailer before a closing time of the retailer. The online concierge system evaluates time intervals by subtracting a travel time for the picker travelling from the retailer to the location from a predicted fulfillment time for the order. This prevents the time for delivering items after being obtained from affecting whether a time interval may be selected.
G06Q 10/087 - Gestion d’inventaires ou de stocks, p.ex. exécution des commandes, approvisionnement ou régularisation par rapport aux commandes
G06Q 10/04 - Prévision ou optimisation spécialement adaptées à des fins administratives ou de gestion, p. ex. programmation linéaire ou "problème d’optimisation des stocks"
An online concierge system receives, from a client device associated with a user of the online concierge system, order data associated with an order placed with the online concierge system, in which the order data describes a delivery location for the order. The online concierge system receives information describing a set of attributes associated with the delivery location and accesses a machine learning model trained to predict a difference between an arrival time and a delivery time for the delivery location. The online concierge system applies the model to the set of attributes associated with the delivery location to predict the difference between the arrival time and the delivery time for the delivery location and determines an estimated delivery time for the order based at least in part on the predicted difference. The online concierge system sends the estimated delivery time for the order for display to the client device.
A keyword campaign automatically groups keywords for customized override bids for the keyword group. The keywords of a campaign may be analyzed by a computer model to predict membership in a category in addition to the likelihood that the bid of the keyword will be modified. The keyword groups may be automatically generated based on the predictions, and performance metrics are evaluated for the keyword groups at one or more modified bids. The performance metrics of the keyword groups at the modified bids may then be used to set override bids. The automatically generated keyword groups and performance metrics permit a sponsor to intelligently group and customize keyword bids with reduced interface interactions and without requiring individual keyword bid adjustments.
An online system generates a knowledge graph database representing relationships between entities in the online system. The online system generates the knowledge graph database by at least obtaining descriptions for an item. The online system generates one or more prompts to a machine-learned language model, where a prompt includes a request to extract a set of attributes for the item from the description of the item. The online system receives a response generated from executing the machine-learned language model on the prompts. The online system parses the response to extract the set of attributes for the item. For each extracted attribute, the online system generates connections between an item node representing the item and a set of attribute nodes for the extracted set of attributes in the database.
A reinforcement learning model selects a content composition based, in part, on inter-session rewards. In addition to near-in-time rewards of user interactions with a content composition for evaluating possible actions, the reinforcement learning model also generates a reward and/or penalty based on between-session information, such as the time between sessions. This permits the reinforcement learning model to learn to evaluate content compositions not only on the immediate user response, but also on the effect of future user engagement. To determine a composition for a search query, the reinforcement learning model generates a state representation of the user and search query and evaluates candidate content compositions based on learned parameters of the reinforcement learning model that evaluates inter-session rewards of the content compositions.
An online concierge system receives information describing one or more interactions with a shared shopping list by at least one of multiple users associated with the shared shopping list and identifies a set of attributes associated with the shared shopping list, in which the set of attributes is based at least in part on the interaction(s). The system accesses a machine learning model trained to predict a time that a user associated with the shared shopping list will place an order including one or more items in the shared shopping list and applies the model to the set of attributes to predict the time. The system generates a notification based at least in part on the time that the user is predicted to place the order and sends the notification to one or more client devices associated with one or more users associated with the shared shopping list.
An online concierge system facilitates creation of shopping lists of items for ordering from a physical retail store and at least partial self-service fulfillment of orders by the customer. To support fulfillment by the customer, the online concierge system may intelligently select one or more items of the order to be picked by a third-party picker and prepopulated to a shopping cart reserved for the customer in advance of the customer arriving at the retail location. The items for prepopulating may be selected based on various factors that optimize prepopulation decisions on an item-by-item basis in accordance with various machine learning models. The online concierge system may furthermore facilitate procurement of the remaining items by the customer through a customer client device that may track item procurement and/or provide guidance for locating items.
An online system performs an inference task in conjunction with the model serving system or the interface system to continuously monitor conversations between requesting users and fulfillment users to determine whether the online system can intervene to automatically respond to a message sent by a sending party, rather than prompting the receiving party for a manual reply. Upon inferring that a message can be automatically responded to, the online system automatically provides a response to the message without the receiving party's manual involvement. The online system can further be augmented to classify and reroute certain requesting user or fulfillment user queries that impact an order's end state by intercepting the conversation on behalf of either party and performing one or more automated actions. If the message is action-oriented, the online system may perform one or more automated actions in response to the message.
An online system receives, from a model serving system, an application programming interface (API) request from a plug-in provided by an online system. The API request includes a list of items obtained from a conversation session of a user with a machine-learned language model application of the model serving system. The online system generates a URL to a landing page for the user for creating a purchase list with the online system based on the list of items. Responsive to receiving a request to access the URL, the online system causes display of the landing page on a client device of the user that displays the purchase list including retailer items for one or more retailers corresponding to the list of items in the API request.
An online concierge system determines whether a user's appeasement request is fraudulent. The online concierge system compares the user's appeasement request rate to the appeasement request rates of similar users in a user segment identified with a user segmentation model. The online concierge system computes an appeasement model that represents the appeasement request rates of the users in the user segment. The online concierge system computes an outlier score for the user based on the appeasement model. The online concierge system compares the outlier score to a threshold. If the outlier score exceeds the threshold, the online concierge system may determine that the appeasement request is not likely fraudulent and thus applies an appeasement action to the user. If the outlier score does not exceed the threshold, the online concierge system may determine that the appeasement request is likely fraudulent and thus applies a security action to the user.
G06Q 30/016 - Fourniture d’une assistance aux clients, p. ex pour assister un client dans un lieu commercial ou par un service d’assistance après-vente
G06F 18/23213 - Techniques non hiérarchiques en utilisant les statistiques ou l'optimisation des fonctions, p.ex. modélisation des fonctions de densité de probabilité avec un nombre fixe de partitions, p.ex. K-moyennes
67.
CLUSTERING DATA DESCRIBING INTERACTIONS PERFORMED AFTER RECEIPT OF A QUERY BASED ON SIMILARITY BETWEEN EMBEDDINGS FOR DIFFERENT QUERIES
An online system leverages stored interactions with items made by users after the online system received queries to determine display of items satisfying the query. For example, the online system trains a model to predict a likelihood of a user performing an interaction with an item displayed after a query was received. As different items receive different amounts of interaction from users, limited historical interaction with certain items may limit accuracy of the model. The online system generates embeddings for previously received queries and uses measures of similarity between embeddings for queries to generate clusters of queries. Previous interactions with queries in a cluster are combined, with the combined data being used for determining display of items in response to a query.
G06F 16/28 - Bases de données caractérisées par leurs modèles, p.ex. des modèles relationnels ou objet
G06F 16/2457 - Traitement des requêtes avec adaptation aux besoins de l’utilisateur
G06F 16/248 - Présentation des résultats de requêtes
G06F 18/2413 - Techniques de classification relatives au modèle de classification, p.ex. approches paramétriques ou non paramétriques basées sur les distances des motifs d'entraînement ou de référence
68.
DISPLAYING AN AUGMENTED REALITY ELEMENT THAT PROVIDES A PERSONALIZED ENHANCED EXPERIENCE AT A WAREHOUSE
An augmented reality application executing on a client device receives video data captured by a camera of the device, in which the video data includes a display area of the device. The application detects a set of items within the display area based on the video data, wherein the set of items is included among an inventory of a warehouse associated with a retailer, and accesses a set of attributes of each item. The application retrieves profile information including a set of preferences associated with a customer of the retailer, matches one or more of the set of preferences with one or more attributes of each item, and generates an augmented reality element based on the matches. The augmented reality element is then displayed, such that it is overlaid onto a portion of the display area based on a location within the display area at which the items are detected.
A system, for example, an online system uses a machine learning based language model, for example, a large language model (LLM) to process high-level natural language queries received from users. The system receives a natural language query from a user of a client device. The system determines contextual information associated with the query. Based on this information, the system generates a prompt for the machine learning based language model. The system receives a response from the machine learning based language model. The system uses the response to generate a search query for a database. The system obtains results returned by the database in response to the search query and provides them to the user. The system allows users to specify high level natural language queries to obtain relevant search results, thereby improving the overall user experience.
A system, for example, an online system uses a machine learning based language model, for example, a large language model (LLM) to process crowd-sourced information provided by users. The crowd-sourced information may include comments from users represented as unstructured text. The system further receives queries from users and answers the queries based on the crowd-sourced information collected by the system. The system generates a prompt for input to a machine-learned language model based on the query. The system provides the prompt to the machine-learned language model for execution and receives a response from the machine-learned language model. The response comprises the insight on the topic and evidence for the insight. The evidence identifies one or more comments used to obtain the insight.
An online system performs inference in conjunction with a machine-learned language model to determine one or more key items in an order. The system generates a prompt for input to a machine-learned language model. The prompt may specify at least the list of ordered items in the order and a request to infer one or more key items in the order. The system provides the prompt to a model serving system for execution by the machine-learned language model for execution. The system parses the response from the model serving system to extract a subset of items as the one or more key items of the order. The system generates an interface presenting the order of the list of items and one or more indications on the interface that indicate the subset of items are key items of the order.
An online system trains a specific-purpose LLM. The online system obtains training examples and divides training examples across batches. The online system generates a specific response by applying parameters of the specific-purpose LLM to a batch of training examples. The online system generates a general response by applying parameters of a general-purpose LLM to the batch of training examples. The online system computes a human readability score representing the difference between the specific response and the general response. The online system computes an objective compliance score by applying an evaluation model to the specific response, the evaluation model trained to score the first response based on a specific objective. The online system updates the parameters of the specific-purpose LLM based on the human readability score and the objective compliance score.
H04L 51/02 - Messagerie d'utilisateur à utilisateur dans des réseaux à commutation de paquets, transmise selon des protocoles de stockage et de retransmission ou en temps réel, p.ex. courriel en utilisant des réactions automatiques ou la délégation par l’utilisateur, p.ex. des réponses automatiques ou des messages générés par un agent conversationnel
73.
SELECTING AN ITEM FOR INCLUSION IN AN ORDER FROM A USER OF AN ONLINE CONCIERGE SYSTEM FROM A GENERIC ITEM DESCRIPTION RECEIVED FROM THE USER
An online concierge system maintains a taxonomy associating one or more specific items offered by a warehouse with a generic item description. When the online concierge system receives a generic item description from a user for inclusion in an order, the online concierge system uses the taxonomy to select a set of items associated with the generic item description. Based on probabilities of the user purchasing various items of the set, the online concierge system selects an item of the set for inclusion in the order For example, the online concierge system selects an item of the set for which the user has a maximum probability of being purchased. Subsequently, the online concierge system displays an interface for the user that is prepopulated with information identifying the selected item of the set.
An online concierge system schedules pickers (shoppers) to fulfill orders from users. During periods of peak demand, the system increases compensation to shoppers to encourage more to participate, thereby reducing missed orders. The system determines an optimal multiplier to increase compensation based on predictive models of supply and demand and then applying an optimization algorithm to search different hyperparameters that affect how the models generate the multipliers. The system selects the optimal multipliers for different time periods and locations. The system may further present the multipliers being offered during future time periods and enable users to activate reminder alerts for select periods. The offers may be presented in a ranked list using a model trained to infer likelihoods of the user accepting participation and/or setting a reminder notification.
An online concierge system facilitates ordering of items by customers, procurement of the items from physical retailers by pickers assigned to the orders, and delivery of the orders to customers. To enable efficient procurement, the online concierge system may facilitate preemptive picking of items for staging at a rapid fulfillment area of the physical retailer, and pickers may selectively pick items from the rapid fulfillment area instead of their standard storage locations. Decisions on which items to preemptively pick may be based on a predictive optimization model that scores and ranks items for predictive picking in accordance with various optimization criteria. In the course of fulfilling orders, pickers may furthermore be assigned to replenish items from the standard storage locations to the rapid fulfillment area to satisfy future predicted or actual orders in a manner that optimizes a cost metric.
G06Q 10/087 - Gestion d’inventaires ou de stocks, p.ex. exécution des commandes, approvisionnement ou régularisation par rapport aux commandes
G06Q 10/04 - Prévision ou optimisation spécialement adaptées à des fins administratives ou de gestion, p. ex. programmation linéaire ou "problème d’optimisation des stocks"
An online concierge system facilitates ordering of items by customers, procurement of the items from physical retailers by pickers assigned to the orders, and delivery of the orders to customers. To enable efficient procurement, the online concierge system may facilitate preemptive picking of items for staging at a rapid fulfillment area of the physical retailer, and pickers may selectively pick items from the rapid fulfillment area instead of their standard storage locations. Decisions on which items to preemptively pick may be based on a predictive optimization model that scores and ranks items for predictive picking in accordance with various optimization criteria. In the course of fulfilling orders, pickers may furthermore be assigned to replenish items from the standard storage locations to the rapid fulfillment area to satisfy future predicted or actual orders in a manner that optimizes a cost metric.
An online system adjusts a guardrail setting used by a user treatment engine based on conditions faced by the online system. The online system simulates the performance of the user treatment engine using different candidate guardrail settings and computes a score for each of the guardrail settings based on the performance of the user treatment engine using each of the guardrail settings. The online system selects a new guardrail setting for the user treatment engine based on the performance scores for the candidate guardrail settings. Furthermore, the online system generates simulated training examples to initially train a user treatment engine. The online system uses a treatment performance model to simulate the effect of treatments applied to users and generates simulated training examples based on the predicted effect of the treatments. The online system retrains the user treatment engine on real training examples that are generated based on actual treatments.
An online system adjusts a guardrail setting used by a user treatment engine based on conditions faced by the online system. The online system simulates the performance of the user treatment engine using different candidate guardrail settings and computes a score for each of the guardrail settings based on the performance of the user treatment engine using each of the guardrail settings. The online system selects a new guardrail setting for the user treatment engine based on the performance scores for the candidate guardrail settings. Furthermore, the online system generates simulated training examples to initially train a user treatment engine. The online system uses a treatment performance model to simulate the effect of treatments applied to users and generates simulated training examples based on the predicted effect of the treatments. The online system retrains the user treatment engine on real training examples that are generated based on actual treatments.
G16H 50/50 - TIC spécialement adaptées au diagnostic médical, à la simulation médicale ou à l’extraction de données médicales; TIC spécialement adaptées à la détection, au suivi ou à la modélisation d’épidémies ou de pandémies pour la simulation ou la modélisation des troubles médicaux
Responsive to an input query from a user, an online system presents a list of recommended items that are related to the input query. The input query may be formulated as a natural language query. The online system performs an inference task in conjunction with the model serving system to generate one or more additional queries that are related to the input query and/or are otherwise related to the recommended items presented in response to the input query. The additional queries may be presented to the user in conjunction with the list of recommended items.
An online system performs an inference task in conjunction with the model serving system to infer one or more purposes of the order of a user that includes a list of ordered items. The model serving system may host a machine-learned language model, and in one instance, the machine-learned language model is a large language model. The online system generates recommendations to the user based on the inferred purpose of the order. The online system may generate one or more recommendations that are equivalent orders having the same or similar purpose as the existing order.
An online system manages campaign participation by a plurality of sub-campaigns with a reinforcement learning model. The reinforcement learning model determines a current context and determines an action that affects the participation of the individual sub-campaigns. The reinforcement learning model may thus dynamically control the participation over time as different objectives are achieved by the sub-campaigns and may account for the different contexts that change over time.
An online concierge system schedules pickers (shoppers) to fulfill orders from users. During periods of peak demand, the system increases compensation to shoppers to encourage more to participate, thereby reducing missed orders. The system determines an optimal multiplier to increase compensation based on predictive models of supply and demand and then applying an optimization algorithm to search different hyperparameters that affect how the models generate the multipliers. The system selects the optimal multipliers for different time periods and locations. The system may further present the multipliers being offered during future time periods and enable users to activate reminder alerts for select periods. The offers may be presented in a ranked list using a model trained to infer likelihoods of the user accepting participation and/or setting a reminder notification.
GENERATING A SUGGESTED SHOPPING LIST BY POPULATING A TEMPLATE SHOPPING LIST OF ITEM CATEGORIES WITH ITEM TYPES AND QUANTITIES BASED ON A SET OF COLLECTION RULES
An online system generates a template shopping list for a user by accessing a machine learning model trained based on historical order information associated with the user, applying the model to predict likelihoods of conversion for item categories by the user, and populating the template shopping list with one or more item categories based on the predicted likelihoods. The system ranks one or more item types associated with each item category in the template shopping list and determines a set of collection rules associated with one or more item categories/types based on the historical order information. The system generates a suggested shopping list by populating each item category in the template shopping list with one or more item types and a quantity of each item type based on the ranking and rules and sends the suggested shopping list and rules for display to a client device associated with the user.
An online concierge system delivers items from multiple retailers to customers. To avoid delivery of expired or near-expired items, the online concierge system obtains attributes of items offered by a retailer, such as from images of items at the retailer from client devices and uses a trained desirability model to predict a desirability score of an item based on the item's attributes. The desirability model is trained using training examples with labels indicating whether an item was suitable for inclusion in an order. The desirability model may be used to determine if an item is suitable for inclusion in an order, to provide suggestions for a retailer for using the item, or to select a retailer for fulfilling an order.
A specific item is identified to suggest a replacement therefor to a user. A set of candidate replacement items for the specific item is determined. For at least one of the candidate replacement items, an expiration score is determined based on expiration information associated with the item. A replacement score for the candidate replacement item is determined by inputting the determined expiration score as a feature into a machine learning model that is trained using features of historical samples of candidate replacement items suggested as a replacement to users and the replacement suggestion being accepted by the users. One or more of the candidate replacement items is selected based on respective replacement scores as one or more suggested replacement items. A graphical user interface of a client device of the user is caused to display the one or more suggested replacement items as the replacement for the specific item.
An online concierge system for establishing a communication session between devices using a light signal. A first client device captures video data depicting a light emitter of another client device. The first client device detects a light signal transmitted by the light emitter in the video data. The first client device extracts a handshake identifier from the light signal by decoding the light signal. A machine learning model may be used to translate the light signal into a numerical or an alphanumerical identifier. The first client device established a communication session with the other client device by transmitting a request to establish the communication session via an online concierge system. The request contains the extracted handshake identifier.
G06K 7/14 - Méthodes ou dispositions pour la lecture de supports d'enregistrement par radiation corpusculaire utilisant la lumière sans sélection des longueurs d'onde, p.ex. lecture de la lumière blanche réfléchie
H04L 15/04 - Appareils ou circuits à l'extrémité d'émission
H04L 67/141 - Configuration des sessions d'application
87.
ONLINE SHOPPING SYSTEM AND METHOD FOR SELECTING A WAREHOUSE FOR INVENTORY BASED ON PREDICTED AVAILABILITY AND PREDICTED REPLACEMENT MACHINE LEARNING MODELS
An online concierge system allows users to order items from a warehouse having multiple physical locations, allowing a user to order items at any given warehouse location. To select a warehouse location for a warehouse selected by a user, the online concierge system identifies a set of items that the user has a threshold likelihood of purchasing from prior orders by the user. For each of a set of warehouse locations, the online concierge system applies a machine-learned item availability model to each item of the identified set. From the availabilities of items of the set at each warehouse location of the set, the online concierge system selects a warehouse location. The online concierge system identifies an inventory of items from the selected warehouse location to the user for inclusion in an order.
An automated checkout system accesses an image of an item inside a shopping cart and receives an identifier determined for the item inside the cart. The automated checkout system determines a load measurement for the item inside the cart using load sensors coupled to the cart. The automated checkout system encodes a feature vector of the item based at least on the determined weight, the accessed image, and the determined identifier. The automated checkout system inputs the feature vector to a machine-learning model to determine a confidence score describing a likelihood that the identifier determined for the item matches the item placed inside the cart. If the confidence score is less than a threshold confidence score, the automated checkout system generates a notification alerting an operator of an anomaly in the identifier.
G06Q 20/40 - Autorisation, p.ex. identification du payeur ou du bénéficiaire, vérification des références du client ou du magasin; Examen et approbation des payeurs, p.ex. contrôle des lignes de crédit ou des listes négatives
B62B 3/14 - Voitures à bras ayant plus d'un essieu portant les roues servant au déplacement; Dispositifs de direction à cet effet; Appareillage à cet effet caractérisées par des moyens pour l'emboîtement ou l'empilage, p.ex. chariots pour achats
B62B 5/00 - Accessoires ou détails spécialement adaptés aux voitures à bras
G06Q 20/20 - Systèmes de réseaux présents sur les points de vente
G06V 10/70 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique
PERSONALIZED RECOMMENDATION OF COMPLEMENTARY ITEMS TO A USER FOR INCLUSION IN AN ORDER FOR FULFILLMENT BY AN ONLINE CONCIERGE SYSTEM BASED ON EMBEDDINGS FOR A USER AND FOR ITEMS
An online concierge shopping system identifies candidate items to a user for inclusion in an order based on prior user inclusion of items in orders and items currently included in the order. From a multi-dimensional tensor generated from cooccurrences of items in orders from various users, the online concierge system generates item embeddings and user embeddings in a common latent space by decomposing the multi-dimensional tensor. From items included in an order, the online concierge system generates an order embedding from item embeddings of the items included in the order. Scores for candidate items are determined based on similarity of item embeddings for the candidate items to the order embedding. Candidate items are selected based on their scores, with the selected candidate items identified to the user.
Systems and methods for a collaborative offer portal is provided. A proposed offer is received from a manufacturer, including an offer structure and a number of consumers they wish to target. Transaction logs of a retailer are accessed to determine an audience for the offer by calculating a return on investment (ROI) for the customer base using the retailer's records given the offer type. The consumers are then grouped by their ROI distribution, and the ROI for the deal is calculated based upon the offer size in light of this distribution. From the offer ROI a discount percentage to be paid by the retailer versus the merchant can be created. The retailer may then choose to accept the offer for deployment.
An online concierge system may receive, from a customer, a selection of an item that is associated with a first brand. The online concierge system may extract features associated with the customer and features associated with the item. The online concierge system may input the extracted features to a machine learning model that is trained to predict a degree of association between the customer and the first brand associated with the item. The online concierge system may identify candidate alternatives for replacing the item. The candidate alternatives may include a first alternative that is associated with the first brand and a second alternative that is associated with a second brand different from the first brand. The online concierge system may select, based on the degree of association between the customer and the first brand, one or more candidate alternatives to be presented to the customer to replace the item.
An online concierge system generates a personalized storefront user interface to recommend items for purchase and delivery to a customer. The online concierge system obtains a user identifier for the customer and generates a set of recommended search terms that it predicts will be relevant to the customer. The recommended search terms may be identified at least in part by mapping items previously purchased by the customer to search queries that resulted in purchases of that item across a population of customers of the online concierge system. The online concierge system then executes respective search queries for the each of the set of search terms to generate respective search result sets for each of the recommended search terms. The search result sets may be presented as respective search queries on a user interface screen of a customer client device.
An online system displays search results in response to a query by receiving a query from a customer. An online system accesses a set of candidate items and computes a relevance score and personalization score for each item. The online system computes the relevance score based on query data and item data and may normalize the relevance score. The online system computes the personalization score based on item data, such as an item embedding, and user data, such as a user embedding. The online system computes a query specificity score and adjusts the personalization score with the query specificity score such that generic queries have high personalization scores and specific queries have low personalization scores. The online system combines the relevance and personalization scores for each candidate item into a ranking score and displays the candidate items to the customer based on their ranking scores.
A method or a system for using machine learning to dynamically boost order delivery time. The system receives an order associated with a delivery time and a compensation value. The system applies a machine-learning model to an order to predict an amount of lateness time that an order will be fulfilled late. The system then determines a lateness value based in part on the predicted amount of lateness time. The lateness value indicates a penalty caused by the predicted amount of lateness time. For each of a plurality of proposed boost amounts for the compensation value, the system determines an uplift, indicating a reduction of the lateness value caused by the boost amount. The system then selects a boost amount from the plurality of boost amounts based in part on the determined uplifts, causing the order to be accepted sooner to thereby boost order delivery time.
A system may generate a prompt based in part on a search query from a customer client device. The prompt instructs a machine learned model to provide item predictions. And the model was trained by: converting structured data describing items of an online catalog to annotated text data (unstructured data), generating training examples based in part on the annotated text data, and training the model using the training examples. The system may receive item predictions generated by the prompt being applied to the machine learned model, the item predictions may have corresponding item identifiers. The item predictions are processed to identify a recommended item from the item predictions. The processing includes determining item information for the recommended item using an item identifier associated with the recommended item. The item information is provided to the customer client device.
An online concierge shopping system fulfills orders using workers who pick items at a warehouse to complete an order and workers to deliver the orders to a customer's location. To optimize the staffing of workers for each task, the system uses a trained model to predict the number of workers needed to achieve an optimal outcome based on an input set of contextual information. The system also schedules specific workers to various shifts using the predicted number of workers needed and then searching a feasibility space for an optimal solution. The trained model may be updated based on performance observations.
An online concierge system generates an aggregated lift score for a test feature for the online concierge system. The online concierge presents prioritized items from a set of item groups to two sets of users: a test set and a control set. The online concierge system uses the test feature to present prioritized items to users in the test set, and the online concierge system uses existing functionality to present prioritized items to users in the control set. For each test group, the online concierge system creates holdout subsets out of the test set and the control set. The online concierge system tracks user interactions with items in an item group and computes a group lift score for the item group. The online concierge system generates an aggregated lift score for the test feature based on the group lift scores and presents items to users based on the aggregated lift score.
An online concierge system delivers items from retailers to customers. The online concierge predicts a range of times during which an order may be fulfilled for presentation to a user. The online concierge system uses a trained maximum time prediction model to determine a maximum time for order fulfillment based on an order. A trained minimum time prediction model determines a minimum time for order fulfillment from the order and the maximum time. The minimum time may account for one or more rules (e.g., a percentage of orders fulfilled before the minimum time, a desired rate of selection of a range including the minimum time). A range bounded by the maximum time and the minimum time is transmitted to a customer to enable the customer to select a time interval for order fulfillment.
An online system uses an offline iterative clustering process to evaluate the performance of a set of content selection frameworks. To perform an iteration of the iterative clustering process, an online system clusters the testing example data into a set of clusters. An online system computes a set of framework scores for each of the generated clusters. An online system computes an improvement score for each cluster based on the performance scores of the clusters. To determine whether to perform another iteration, an online system computes an aggregated improvement score based on the improvement scores of the clusters. If an online system determines that the aggregated improvement score does not meet the threshold, an online system performs another iteration of the process above. When an online system finishes the iterative process, an online system outputs the improvement scores of the most-recent iteration.
A system accesses user data describing characteristics of a user and generates a content item score for each content item of a plurality of content items. The system generates the content item score by applying a machine-learning model to the user data, and then generates a plurality of content bundles. The system also generates a bundle score for each content bundle based on corresponding content item scores for the content item associated with each content bundle, randomly selects a bundle of the plurality of content bundles based on the generated bundle scores, and transmits the randomly selected bundle to a client device associated with the user for display to the user. Finally, the system applies the model to each of the generated training examples and updates the parameters of the model based on the model output.
brevets
