Predictive recommendation system using tiered feature data
AssigneeGroupon, Inc.
InventorLawrence Lee Wai
Life of this patent
Estimating the term…
Abstract
In general, embodiments of the present invention provide systems, methods and computer readable media for a predictive recommendation system using predictive models derived from tiered feature data.
Claims17
A computer-implemented method for generating a combined feature representing a promotion, comprising:
- receiving DDO feature data representing the promotion, wherein the DDO feature data include data collected from a first set of impressions associated with the promotion permalink;
- determining, using a processor, whether the first set of impressions satisfies a threshold quantity of impressions;
- and
- in an instance in which the first set of impressions does not satisfy the threshold quantity,
- generating tiered DDO feature data representing the promotion, wherein the tiered DDO feature data were collected from a second set of impressions representing a tier group of promotions including the promotion;
- receiving historical feature data representing the tier group of promotions;
- integrating the tiered DDO feature data and the historical feature data;
- generating the combined feature representing the promotion using the integrated tiered DDO feature data and the historical feature data;
- selecting, based on the combined feature, a subset of promotions from a plurality of promotions, the subset of promotions to be recommended to a customer;
- and
- transmitting the subset of promotions to a client device associated with the customer for display via the client device.
An apparatus, operable on a distributed computing system, to generate a combined feature representing a promotion, the apparatus comprising:
- a relevance engine, wherein the relevance engine is stored in a memory such that when executed by a processor the relevance engine is configured to receive one or more user requests and output one or more ranked available promotions in response to the user request, the ranked available promotions generated in part by at least a predictive model builder and a DDO feature data integrator;
- wherein the predictive model builder is stored in the memory such that when executed by a processor the predictive model builder is configured to integrate DDO feature data, received via a network interface representing the promotion permalink and historical feature data representing the promotion permalink, wherein the DDO feature data include data collected from a first set of impressions associated with the promotion permalink;
- and
- wherein the DDO feature data integrator is stored in the memory such that when executed by a processor the DDO feature data integrator is configured to determine that the first set of impressions satisfies a threshold quantity of impressions and to generate a combined feature representing the promotion using the integrated DDO feature data and the historical feature data;
- and
- wherein the relevance engine is configured to select, based on the combined feature, a subset of promotions from a plurality of promotions, the subset of promotions to be recommended to a customer;
- and
- a communications module configured to transmit the subset of promotions to a client device associated with the customer for display via the client device.
A computer-implemented method for generating a combined feature representing a promotion, comprising:
- receiving DDO feature data representing the promotion, wherein the DDO feature data include data collected from a first set of impressions associated with the promotion permalink integrating the DDO feature data and the historical feature data;
- generating a combined feature representing the promotion using the integrated DDO feature data and the historical feature data;
- selecting, based on the combined feature, a subset of promotions from a plurality of promotions, the subset of promotions to be recommended to a customer;
- and
- transmitting the subset of promotions to a client device associated with the customer for display via the client device.
A computer program product for generating one or more user selections for display as an impression via a promotion and marketing service, the computer program product comprising a non-transitory computer-readable storage medium storing computer-readable program instructions that, when executed, cause at least one computer to perform operations comprising:
- receiving DDO feature data representing the promotion, wherein the DDO feature data include data collected from a first set of impressions associated with the promotion permalink;
- determining, using a processor, whether the first set of impressions satisfies a threshold quantity of impressions;
- and
- in an instance in which the first set of impressions does not satisfy the threshold quantity,
- generating tiered DDO feature data representing the promotion, wherein the tiered DDO feature data were collected from a second set of impressions representing a tier group of promotions including the promotion;
- receiving historical feature data representing the tier group of promotions;
- integrating the tiered DDO feature data and the historical feature data;
- generating the combined feature representing the promotion using the integrated tiered DDO feature data and the historical feature data;
- selecting, based on the combined feature, a subset of promotions from a plurality of promotions, the subset of promotions to be recommended to a customer;
- and
- transmitting the subset of promotions to a client device associated with the customer for display via the client device.
Dependencies are inferred from the claim text.