Abstract: Example order management systems and methods are described. In one implementation, a catalog ingestion system receives data associated with multiple products offered by multiple brands. A commerce management system receives a customer order from a customer. The customer order includes at least a portion of the multiple products offered by the multiple brands. The commerce management system splits the customer order into multiple brand orders such that each brand order is associated with products in the customer order from a particular brand. An integration platform fulfills the brand orders by submitting the brand orders to the associated brands.

Abstract: Systems and methods for optimizing the combination of products and services a business offers to customers, identifying a combination of top markets a business offers to customers for growth opportunities, and optimizing advertisements. In one implementation, the disclosed system includes at least one processing device and at least one non-transitory memory containing software code configured to cause the processing device to: gather customer data and financial institution data from a plurality of data sources; extract a plurality of customer behavior features and a plurality of financial institution behavior features; process the customer behavior features and financial institution behavior features using one or more trained foundation models; input the foundation model outputs and a plurality of goal inputs into a trained product model; and output a natural-language product response.

Abstract: Systems and methods for optimizing the combination of products and services a business offers to customers, identifying a combination of top markets a business offers to customers for growth opportunities, and optimizing advertisements. In one implementation, the disclosed system includes at least one processing device and at least one non-transitory memory containing software code configured to cause the processing device to: gather customer data and financial institution data from a plurality of data sources; extract a plurality of customer behavior features and a plurality of financial institution behavior features; process the customer behavior features and financial institution behavior features using one or more trained foundation models; input the foundation model outputs and a plurality of goal inputs into a trained product model; and output a natural-language advertisement response.

Abstract: Systems and methods for optimizing the combination of products and services a business offers to customers, identifying a combination of top markets a business offers to customers for growth opportunities, and optimizing advertisements. In one implementation, the disclosed system includes at least one processing device and at least one non-transitory memory containing software code configured to cause the processing device to: gather customer data and financial institution data from a plurality of data sources; extract a plurality of customer behavior features and a plurality of financial institution behavior features; process the customer behavior features and financial institution behavior features using one or more trained foundation models; input the foundation model outputs and a plurality of goal inputs into a trained product model; and output a natural-language market response.

Abstract: Query language statements are generated from natural language statements using a knowledge graph representing one or more databases. The knowledge graph is obtained by creating nodes representing tables and operations referenced by queries to the databases. The data of the databases is evaluated to identify entities and dimensions of entities from among the nodes. The entities are assigned human-understandable labels by an LLM. A natural language statement is converted to a knowledge graph language (KGL) statement and references in the KGL statement are replaced with references to entities in the knowledge graph. The KGL statement is then programmatically converted to a database language statement.

Abstract: A user preference hierarchy is determined from user response to images. Images may be tagged using machine learning models trained to determine values for images. Products are clustered according to product vectors. Images of products within a cluster are clustered according to composition and groups of images are selected from image clusters for soliciting feedback regarding user preference for products of a cluster. Feedback is used to train a user preference model to estimate affinity for a product vector. A user may provide feedback regarding a price point and products are weighted according to a distribution about the price point. The distribution may be asymmetrical according to direction of movement of the price point. Filters may be dynamically defined and presented to a user based on popularity and frequency of occurrence of attribute-value pairs of search results and based on feedback regarding the search results.

Abstract: Query language statements are generated from natural language statements using a knowledge graph representing one or more databases. The knowledge graph is obtained by creating nodes representing tables and operations referenced by queries to the databases. The data of the databases is evaluated to identify entities and dimensions of entities from among the nodes. The entities are assigned human-understandable labels by an LLM. A natural language statement is converted to a knowledge graph language (KGL) statement and references in the KGL statement are replaced with references to entities in the knowledge graph. The KGL statement is then programmatically converted to a database language statement.

Abstract: A user preference hierarchy is determined from user response to images. Images may be tagged using machine learning models trained to determine values for images. Products are clustered according to product vectors. Images of products within a cluster are clustered according to composition and groups of images are selected from image clusters for soliciting feedback regarding user preference for products of a cluster. Feedback is used to train a user preference model to estimate affinity for a product vector. A user may provide feedback regarding a price point and products are weighted according to a distribution about the price point. The distribution may be asymmetrical according to direction of movement of the price point. Filters may be dynamically defined and presented to a user based on popularity and frequency of occurrence of attribute-value pairs of search results and based on feedback regarding the search results.

Abstract: Query language statements are generated from natural language statements using a knowledge graph representing one or more databases. An LLM may be used to generate database language statements from natural language statements. The database language statements may be modified based on the knowledge graph. The database language statements may be corrected using a correction database. The correction database may include entries including a natural language statement, an original database language statement, and one or more corrections. Entries may be corrected in response to human corrections of outputs of the LLM. Entries with common corrections may be consolidated.

Abstract: A traffic-and-billing system imports a chronological schedule from a media automation system. The schedule includes media and advertisement spots allocated in accordance with a content allocation prediction, which allocates a portion of the advertisement spots to general content and a portion to local content. The traffic-and-billing system stores information associated with general advertisement content and generates a content listing based on the content allocation prediction. A general advertisement system receives, an update request that indicates at least one item of the general advertisement content is to be removed from the content listing. The general advertisement system plays out general advertisement content items included in the general advertisement content listing in response to receiving a playout request from the media automation system.

Abstract: Systems and methods are presented for managing the cadence (e.g., frequency or rate) that electronic promotion correspondence is sent to a consumer. A system may access a target cadence indicator associated with a consumer that indicates of a target rate for sending electronic promotion correspondence to the consumer. The system may also determine an actual cadence indicator for the consumer over a predetermined period of time and analyze a potential electronic promotion correspondence for sending to the consumer. The system determines whether to send the electronic correspondence to the consumer based on the target cadence indicator, the actual cadence indicator, and the analysis of the electronic promotion correspondence.

Abstract: Query language statements are generated from natural language statements using a knowledge graph representing one or more databases. The knowledge graph is obtained by creating nodes representing tables and operations referenced by queries to the databases. The data of the databases is evaluated to identify entities and dimensions of entities from among the nodes. The entities are assigned human-understandable labels by an LLM. A natural language statement is converted to a knowledge graph language (KGL) statement and references in the KGL statement are replaced with references to entities in the knowledge graph. The KGL statement is then programmatically converted to a database language statement.

Abstract: Query language statements are generated from natural language statements using a knowledge graph representing one or more databases. The knowledge graph is obtained by creating nodes representing tables and operations referenced by queries to the databases. The data of the databases is evaluated to identify entities and dimensions of entities from among the nodes. The entities are assigned human-understandable labels by an LLM. A natural language statement is converted to a knowledge graph language (KGL) statement and references in the KGL statement are replaced with references to entities in the knowledge graph. The KGL statement is then programmatically converted to a database language statement.

Abstract: In a promotion offering system, a promotion offering may be presented to a consumer on more than one occasion. The previous presentation of the promotion to the consumer may affect a subsequent presentation of the same promotion to the same consumer. The present invention provides an apparatus and method for analyzing the effect a previous presentation of a promotion may have on a consumer when the promotion is presented to the consumer at a subsequent time.

Abstract: Example order management systems and methods are described. In one implementation, a catalog ingestion system receives data associated with multiple products offered by multiple brands. A commerce management system receives a customer order from a customer. The customer order includes at least a portion of the multiple products offered by the multiple brands. The commerce management system splits the customer order into multiple brand orders such that each brand order is associated with products in the customer order from a particular brand. An integration platform fulfills the brand orders by submitting the brand orders to the associated brands.

Abstract: A computer system extracts product data from a website and correlates product records from multiple sources to one another as corresponding to the same product. A website is crawled efficiently by rendering webpages using a virtual browser that ignores blacklisted elements, extracts data from objects without rendering, and suppressing retrieval of remote resources. Data is extracted according to engine control statements including a selector and extractor. A website may be crawled repeatedly and changes in extracted data may be detected and flagged. Engine control statements may be automatically changed in response to detecting a change in the configuration of the website. Images of product records may be correlated with one another by first comparing text of the product records and selecting images for comparison based on composition. Images are compared using a machine learning model. Images determined to be similar may be presented to a human for a correlation decision.

Abstract: Example order management systems and methods are described. In one implementation, a catalog ingestion system receives data associated with multiple products offered by multiple brands. A commerce management system receives a customer order from a customer. The customer order includes at least a portion of the multiple products offered by the multiple brands. The commerce management system splits the customer order into multiple brand orders such that each brand order is associated with products in the customer order from a particular brand. An integration platform fulfills the brand orders by submitting the brand orders to the associated brands.

Abstract: A user preference hierarchy is determined from user response to images. Images may be tagged using machine learning models trained to determine values for images. Products are clustered according to product vectors. Images of products within a cluster are clustered according to composition and groups of images are selected from image clusters for soliciting feedback regarding user preference for products of a cluster. Feedback is used to train a user preference model to estimate affinity for a product vector. A user may provide feedback regarding a price point and products are weighted according to a distribution about the price point. The distribution may be asymmetrical according to direction of movement of the price point. Filters may be dynamically defined and presented to a user based on popularity and frequency of occurrence of attribute-value pairs of search results and based on feedback regarding the search results.

Abstract: A promotion offering system and method is disclosed. The promotion offering system and method selects consumers for a test promotion that has one or more attributes, and is configured to generate test data for multiple consumer groupings. The promotion offering system and method may use the test data in analyzing whether to send a promotion to a consumer. For example, the promotion offering system and method may use one algorithm to generate a list of ranked promotions, and may use the test data in order to adjust the list of ranked promotions (such as replacing a highest ranked promotion with another promotion).

Abstract: A user preference hierarchy is determined from user response to images. Images may be tagged using machine learning models trained to determine values for images. Products are clustered according to product vectors. Images of products within a cluster are clustered according to composition and groups of images are selected from image clusters for soliciting feedback regarding user preference for products of a cluster. Feedback is used to train a user preference model to estimate affinity for a product vector. A user may provide feedback regarding a price point and products are weighted according to a distribution about the price point. The distribution may be asymmetrical according to direction of movement of the price point. Filters may be dynamically defined and presented to a user based on popularity and frequency of occurrence of attribute-value pairs of search results and based on feedback regarding the search results.