Back to results
US 11,961,119 B2Granted
Archive offer personalization
AssigneeScribd, Inc.
InventorJames August Hill
Life of this patent
Estimating the term…
Abstract
Various embodiments of an apparatus, method(s), system(s) and a computer program product(s) described herein are directed to a Offer Engine. The Offer Engine extracts one or more features from data associated with a first user requesting access to a portion of content of a content corpus. The Offer Engine feeds at least one of the features of the first user into a decision tree. The decision tree has multiple levels, wherein at least one level comprises a plurality of leaves and each respective leaf implements at least one machine learning model. The Offer Engine determines whether to provide the first user with a subscription fee offer first option or a non-subscription fee offer second option based at least in part on output of the decision tree.
Claims17
A computer-implemented method, comprising:
- generating training data based on a uniform distribution of users sampled from a user segment;
- training respective machine learning models that correspond to leaves of a plurality of levels of a decision tree, each respective leaf associated with a corresponding confidence interval, the confidence interval representing an expected accuracy range of leaf output;
- extracting one or more features from request data associated with a first user requesting access to user content;
- feeding at least one of the features of the first user into the decision tree;
- identifying, at a first level of the decision tree, a user segment relevant to the first user due to one or more of the extracted features;
- identifying a relevant leaf, at a subsequent level of the decision tree below the first level, that corresponds the to the user segment relevant to the first user;
- generating, at the relevant leaf, output comprising:
- (a) a first prediction of receiving at least one subsequent content access request from the first user;
- and
- (b) a second prediction of acceptance of a first type of offer by the first user in exchange for granting the subsequent content access request;
- feeding the first and the second predictions to an additional leaf at a further subsequent level of the decision tree, the further subsequent level situated below the subsequent level of the relevant leaf;
- generating, at the additional leaf, output comprising:
- (i) a third prediction of receiving at least one content upload from the first user;
- and
- (ii) a fourth prediction of acceptance of a second type of offer by the first user in exchange for the at least one content upload;
- determining whether to provide the first user with the first type of offer or the second type of offer based on the respective output predictions and further based on the confidence intervals of the relevant leaf and the additional leaf.
13 more elements
A system comprising one or more processors, and a non-transitory computer-readable medium including one or more sequences of instructions that, when executed by the one or more processors, cause the system to perform operations comprising:
- generating training data based on a uniform distribution of users sampled from a user segment;
- training respective machine learning models that correspond to leaves of a plurality of levels of a decision tree, each respective leaf associated with a corresponding confidence interval, the confidence interval representing an expected accuracy range of leaf output;
- extracting one or more features from request data associated with a first user requesting access to user content;
- feeding at least one of the features of the first user into the decision tree;
- identifying, at a first level of the decision tree, a user segment relevant to the first user due to one or more of the extracted features;
- identifying a relevant leaf, at a subsequent level of the decision tree below the first level, that corresponds the to the user segment relevant to the first user;
- generating, at the relevant leaf, output comprising:
- (a) a first prediction of receiving at least one subsequent content access request from the first user;
- and
- (b) a second prediction of acceptance of a first type of offer by the first user in exchange for granting the subsequent content access request;
- feeding the first and the second predictions to an additional leaf at a further subsequent level of the decision tree, the further subsequent level situated below the subsequent level of the relevant leaf;
- generating, at the additional leaf, output comprising:
- (i) a third prediction of receiving at least one content upload from the first user;
- and
- (ii) a fourth prediction of acceptance of a second type of offer by the first user in exchange for the at least one content upload;
- determining whether to provide the first user with the first type of offer or the second type of offer based on the respective output predictions and further based on the confidence intervals of the relevant leaf and the additional leaf.
13 more elements
A computer program product comprising a non-transitory computer-readable medium having a computer-readable program code embodied therein to be executed by one or more processors, the program code including instructions for:
- generating training data based on a uniform distribution of users sampled from a user segment;
- training respective machine learning models that correspond to leaves of a plurality of levels of a decision tree, each respective leaf associated with a corresponding confidence interval, the confidence interval representing an expected accuracy range of leaf output;
- extracting one or more features from request data associated with a first user requesting access to user content;
- feeding at least one of the features of the first user into the decision tree;
- identifying, at a first level of the decision tree, a user segment relevant to the first user due to one or more of the extracted features;
- identifying a relevant leaf, at a subsequent level of the decision tree below the first level, that corresponds the to the user segment relevant to the first user;
- generating, at the relevant leaf, output comprising:
- (a) a first prediction of receiving at least one subsequent content access request from the first user;
- and
- (b) a second prediction of acceptance of a first type of offer by the first user in exchange for granting the subsequent content access request;
- feeding the first and the second predictions to an additional leaf at a further subsequent level of the decision tree, the further subsequent level situated below the subsequent level of the relevant leaf;
- generating, at the additional leaf, output comprising:
- (i) a third prediction of receiving at least one content upload from the first user;
- and
- (ii) a fourth prediction of acceptance of a second type of offer by the first user in exchange for the at least one content upload;
- determining whether to provide the first user with the first type of offer or the second type of offer based on the respective output predictions and further based on the confidence intervals of the relevant leaf and the additional leaf.
13 more elements
Dependencies are inferred from the claim text.