# US 2019/0318390 A1 · Predictive Recommendation System Using Contextual Relevance > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2019/0318390 A1 - Document ID: US20190318390A1 - Status: Published application - Kind: A1 - Application number: 16/238,468 - Filed: 2019-01-02 - Published: 2019-10-17 - Earliest claimed: 2013-11-25 - Assignees: Groupon, Inc. - Applicants: Groupon |nc. - Inventors: Lawrence Lee Wai - Web page: https://patentgrep.com/patent/US20190318390A1 ## CPC - G06Q 30/0269: {based on user profile or attribute} - G06Q 30/0207: Discounts or incentives, e.g. coupons or rebates - G06Q 30/0246: {Traffic} - G06Q 30/0247: {Calculate past, present or future revenues} - G06Q 30/0251: Targeted advertisements - G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR ## Abstract In general, embodiments of the present invention provide systems, methods and computer readable media for ranking promotions selected for recommendation to consumers based on predictions of promotion performance and consumer behavior. In embodiments, a set of promotions to be recommended to a consumer can be sorted and/or ranked according to respective relevance scores representing a probability that the consumer's behavior in response to the promotion will match a ranking target. In embodiments, calculating scores is based on a relevance model (a predictive function) derived from one or more contextual data sources representing attributes of promotions and consumer behavior. In embodiments, an absolute relevance score represents an absolute prediction of a ranking target variable. In embodiments, absolute relevance may be used to determine personalized local merchant discovery frontiers; featured result set thresholding for impressions; and/or promotion notification triggers. In embodiments, predictive models based on gross revenue may be optimized using promotion category-dependent price boosting. ## Claims ### Claim at position 1 1-24. (canceled) ### Claim 25 A computer-implemented method, comprising: receiving, from a computing device associated with a consumer, input data representing a request for available promotions; receiving user-promotion interaction data describing attributes of past interactions between the consumer and each available promotion of a set of available promotions; calculating a contextual relevance score for each available promotion of the set of available promotions using the input data and the user-promotion interaction data, wherein the calculating is based on at least one relevance model; generating a subset of available promotions from the set of available promotions, wherein the subset of available promotions comprises available promotions with highest contextual relevance scores among contextual relevance scores calculated for the set of available promotions; and transmitting the subset of available promotions to the computing device. ### Claim 26 (depends on 25) The computer-implemented method of claim 25, further comprising: ranking the subset of available promotions based on their respective contextual relevance scores. ### Claim 27 (depends on 25) The computer-implemented method of claim 25, wherein the relevance model is a predictive function, and calculating the contextual relevance score for each available promotion of the set of available promotions comprises: extracting promotion freshness features representing at least one of the input data and the user-promotion interaction data; generating a promotion freshness feature vector representing a set of extracted promotion freshness features; mapping the promotion freshness feature vector, using the predictive function, to a probability class of a set of probability classes, each probability class representing a different probability that the promotion freshness feature vector matches a ranking target; and calculating the contextual relevance score using the probability class to which the promotion freshness feature vector was mapped. ### Claim 28 (depends on 27) The computer-implemented method of claim 27, wherein the predictive function is generated offline using machine learning. ### Claim 29 (depends on 28) The computer-implemented method of claim 28, wherein the predictive function is generated based in part on at least one contextual data source, wherein the contextual data source is a prior consumer location, a current consumer location, a potential consumer location, categories of interest associated with the prior consumer location, categories of interest associated with the current consumer location, categories of interest associated with the potential consumer location, a prior promotion interest level shown by the consumer, a prior promotion subcategory interest level shown by the consumer, prior promotion clicks made by the consumer, prior promotion views made by the consumer, prior promotion positive feedbacks made by the consumer, or prior promotion purchases made by the consumer. ### Claim 30 (depends on 27) The computer-implemented method of claim 27, wherein the predictive function is an ensemble of trees, a clustering algorithm, a classifier, a neural network, or a random forest algorithm. ### Claim 31 (depends on 25) The computer-implemented method of claim 25, wherein calculating the contextual relevance score for each available promotion of the set of available promotions comprises: extracting promotion freshness features representing at least one of the input data and the user-promotion interaction data; determining an activation state of the computing device associated with the consumer; and in circumstances where the computing device is in an active state, calculating the contextual relevance score for each available promotion of the set of available promotions using the promotion freshness features and a gross revenue relevance model. ### Claim 32 (depends on 31) The computer-implemented method of claim 31, further comprising: in circumstances where the computing device is in an inactivate state, calculating the contextual relevance score for each available promotion of the set of available promotions using the promotion freshness features and a conversion rate relevance model. ### Claim 33 A computer program product, encoded on a computer-readable medium, operable to cause data processing apparatus to perform operations comprising: receiving, from a computing device associated with a consumer, input data representing a request for available promotions; receiving user-promotion interaction data describing attributes of past interactions between the consumer and each available promotion of a set of available promotions; calculating a contextual relevance score for each available promotion of the set of available promotions using the input data and the user-promotion interaction data, wherein the calculating is based on at least one relevance model; generating a subset of available promotions from the set of available promotions, wherein the subset of available promotions comprises available promotions with highest contextual relevance scores among contextual relevance scores calculated for the set of available promotions; and transmitting the subset of available promotions to the computing device. ### Claim 34 (depends on 33) The computer program product of claim 33, further comprising: ranking the subset of available promotions based on their respective contextual relevance scores. ### Claim 35 (depends on 33) The computer program product of claim 33, wherein the relevance model is a predictive function, and calculating the contextual relevance score for each available promotion of the set of available promotions comprises: extracting promotion freshness features representing at least one of the input data and the user-promotion interaction data; generating a promotion freshness feature vector representing a set of extracted promotion freshness features; mapping the promotion freshness feature vector, using the predictive function, to a probability class of a set of probability classes, each probability class representing a different probability that the promotion freshness feature vector matches a ranking target; and calculating the contextual relevance score using the probability class to which the promotion freshness feature vector was mapped. ### Claim 36 (depends on 35) The computer program product of claim 35, wherein the predictive function is generated offline using machine learning. ### Claim 37 (depends on 36) The computer program product of claim 36, wherein the predictive function is generated based in part on at least one contextual data source, wherein the contextual data source is a prior consumer location, a current consumer location, a potential consumer location, categories of interest associated with the prior consumer location, categories of interest associated with the current consumer location, categories of interest associated with the potential consumer location, a prior promotion interest level shown by the consumer, a prior promotion subcategory interest level shown by the consumer, prior promotion clicks made by the consumer, prior promotion views made by the consumer, prior promotion positive feedbacks made by the consumer, or prior promotion purchases made by the consumer. ### Claim 38 (depends on 35) The computer program product of claim 35, wherein the predictive function is an ensemble of trees, a clustering algorithm, a classifier, a neural network, or a random forest algorithm. ### Claim 39 (depends on 33) The computer program product of claim 33, wherein calculating the contextual relevance score for each available promotion of the set of available promotions comprises: extracting promotion freshness features representing at least one of the input data and the user-promotion interaction data; determining an activation state of the computing device associated with the consumer; and in circumstances where the computing device is in an active state, calculating the contextual relevance score for each available promotion of the set of available promotions using the promotion freshness features and a gross revenue relevance model. ### Claim 40 (depends on 39) The computer program product of claim 39, further comprising: in circumstances where the computing device is in an inactivate state, calculating the contextual relevance score for each available promotion of the set of available promotions using the promotion freshness features and a conversion rate relevance model. ### Claim 41 A system comprising one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations, comprising: receiving, from a computing device associated with a consumer, input data representing a request for available promotions; receiving user-promotion interaction data describing attributes of past interactions between the consumer and each available promotion of a set of available promotions; calculating a contextual relevance score for each available promotion of the set of available promotions using the input data and the user-promotion interaction data, wherein the calculating is based on at least one relevance model; generating a subset of available promotions from the set of available promotions, wherein the subset of available promotions comprises available promotions with highest contextual relevance scores among contextual relevance scores calculated for the set of available promotions; and transmitting the subset of available promotions to the computing device. ### Claim 42 (depends on 41) The system of claim 41, further comprising: ranking the subset of available promotions based on their respective contextual relevance scores. ### Claim 43 (depends on 41) The system of claim 41, wherein the relevance model is a predictive function, and calculating the contextual relevance score for each available promotion of the set of available promotions comprises: extracting promotion freshness features representing at least one of the input data and the user-promotion interaction data; generating a promotion freshness feature vector representing a set of extracted promotion freshness features; mapping the promotion freshness feature vector, using the predictive function, to a probability class of a set of probability classes, each probability class representing a different probability that the promotion freshness feature vector matches a ranking target; and calculating the contextual relevance score using the probability class to which the promotion freshness feature vector was mapped. ### Claim 44 (depends on 43) The system of claim 43, wherein the predictive function is generated offline using machine learning. ### Claim 45 (depends on 44) The system of claim 44, wherein the predictive function is generated based in part on at least one contextual data source, wherein the contextual data source is a prior consumer location, a current consumer location, a potential consumer location, categories of interest associated with the prior consumer location, categories of interest associated with the current consumer location, categories of interest associated with the potential consumer location, a prior promotion interest level shown by the consumer, a prior promotion subcategory interest level shown by the consumer, prior promotion clicks made by the consumer, prior promotion views made by the consumer, prior promotion positive feedbacks made by the consumer, or prior promotion purchases made by the consumer. ### Claim 46 (depends on 43) The system of claim 43, wherein the predictive function is an ensemble of trees, a clustering algorithm, a classifier, a neural network, or a random forest algorithm. ### Claim 47 (depends on 41) The system of claim 41, wherein calculating the contextual relevance score for each available promotion of the set of available promotions comprises: extracting promotion freshness features representing at least one of the input data and the user-promotion interaction data; determining an activation state of the computing device associated with the consumer; and in circumstances where the computing device is in an active state, calculating the contextual relevance score for each available promotion of the set of available promotions using the promotion freshness features and a gross revenue relevance model. ### Claim 48 (depends on 47) The system of claim 47, further comprising: in circumstances where the computing device is in an inactivate state, calculating the contextual relevance score for each available promotion of the set of available promotions using the promotion freshness features and a conversion rate relevance model.