Customer data verification in management and determination of user identity using identity graphs
A user identity management platform is provided that manages user identity for an enterprise, such as a retail enterprise. In particular, a specific identity graph structure is provided that allows for flexible management and selection of user account information depending on the context in which that user account information is to be used. Confidence scores may be maintained for nodes and edges and probabilistic techniques associated with account activity may be used to improve confidence of association among nodes within a cluster representing a unique user.
1 . A method of verifying customer data for inclusion in a customer identity graph, the method being performed by a customer identity management platform of a retail enterprise, the method comprising:
obtaining transaction data associated with a third party payment card of a customer used in a transaction at a location of a retail enterprise, the transaction data including transaction attributes including at least a customer name, an account identifier of the third party payment card, and the location of the retail enterprise at which the transaction occurred;
submitting at least a portion of the transaction data to a third party service after completion of a transaction represented by the transaction data, wherein the third party service is external to the retail enterprise, and wherein submitting a portion of the transaction data includes deidentifying the transaction data associated with the third party payment card;
receiving, from the third party service, a plurality of potential customer contact information data sets of the customer, in response to submitting the at least a portion of the transaction data to the third party service, the plurality of potential customer contact information data sets including customer addresses, emails, and phone numbers;
identifying, within the customer identity graph maintained by the identity management platform, a plurality of user profile nodes corresponding to the plurality of potential customer contact information data sets of the customer, the customer identity graph including a plurality of user clusters, each user cluster being associated with a different customer and including one or more user profile nodes, each of the one or more user profile nodes being associated with a customer account of the customer, wherein the identity management platform further includes a random forest classifier model trained using transaction data and account to transaction correlations representing pairs of known correlated nodes or pairs of known non-correlated nodes;
comparing, via the random forest classifier model, the transaction data to other transaction data associated with the plurality of profile nodes, wherein the identity management platform further includes the random forest classifier model;
generating, via the classifier model, based on the comparison, a probability score, the probability score representing a likelihood of correlation between the transaction data and transaction data of a corresponding one of the plurality of user profile nodes;
identifying, based on the probability score generated by the classifier model, a matching user profile node from among the plurality of user profile nodes based on a similarity of the transaction data; and
associating the transaction data with the matching user profile node based on the probability score, wherein associating the transaction data with the matching user profile node further comprises at least one of updating an existing edge confidence, creating a new edge, or removing an existing edge to improve the accuracy and completeness of the customer identity graph,
wherein updating the existing edge confidence includes adjusting a strength of association between the matching user profile node to one or more user profile nodes within the customer identity graph,
wherein creating the new edge includes, in response to determining that the probability score is above an upper identity edge threshold, creating the new edge between the matching user profile node and one or more user profile nodes within the customer identity graph, and
wherein removing the existing edge includes, in response to determining that a probability score is below a lower identity edge threshold, removing the existing edge between the matching user profile node and one or more user profile nodes within the customer identity graph;
receiving a request for a customer identity of the customer, the request comprising attributes of the customer and a desired identity confidence;
in response to receiving the request, determining a user cluster from among the plurality of user clusters corresponding to the customer identity; and
identifying the user profile nodes associated with the user cluster node based on the received attributes and the desired identity confidence, wherein identifying the user profile nodes further comprises identifying one or more customer accounts associated with the user profile nodes within the determined user cluster that satisfy the desired identity confidence based on a node confidence of the user profile nodes corresponding to the one or more customer accounts; and
providing an identification of the one or more customer accounts identified by the user profile nodes within the determined user cluster.
2 . The method of claim 1 , wherein the comparison of the transaction data to other transaction data associated with the plurality of user profile nodes includes an identification of a common item preference between items represented in the transaction data and the other transaction data.
3 . The method of claim 1 , wherein the transaction data further includes a card number of the third party payment card.
4 . The method of claim 3 , further comprising submitting at least a portion of the transaction data to the third party service by providing the customer name, the location of the retail enterprise at which the transaction occurred, and a deidentified representation of the card number.
5 . The method of claim 1 , wherein updating a node confidence of the matching user profile node includes upgrading the node confidence from a low node confidence to a higher node confidence.
6 . The method of claim 1 , wherein the third party service is a data broker service.
7 . A customer identity management platform used within a retail enterprise, the customer identity management platform comprising:
a computing system comprising a memory and a processor, the memory storing instructions which, when executed by the processor, cause the computing system to perform:
obtaining transaction data associated with a third party payment card of a customer used in a transaction at a location of a retail enterprise location, the transaction data including transaction attributes including at least a customer name, an account identifier of the third party payment card, and the location of the retail enterprise at which the transaction occurred;
submitting at least a portion of the transaction data to a third party service after completion of a transaction represented by the transaction data, wherein the third party service is external to the retail enterprise, and wherein submitting a portion of the transaction data includes deidentifying the transaction data associated with the third party payment card;
receiving, from the third party service, a plurality of potential customer contact information data sets of the customer in response to submitting the at least a portion of the transaction data to the third party service, the plurality of potential customer contact information data sets including customer addresses, emails, and phone numbers;
identifying, within an identity graph maintained by the customer identity management platform, a plurality of user profile nodes corresponding to the plurality of potential customer contact information data sets of the customer, the customer identity graph including a plurality of user clusters, each user cluster being associated with a different customer and including one or more user profile nodes, each of the one or more user profile nodes being associated with a customer account of the customer, wherein the identity management platform further includes a random forest classifier model trained using transaction data and account to transaction correlations representing pairs of known correlated nodes or pairs of known non-correlated nodes;
comparing, via the random forest classifier model, the transaction data to other transaction data associated with the plurality of profile nodes, wherein the identity management platform further includes the random forest classifier model;
generating, via the classifier model, based on the comparison, a correlation probability score, the probability score representing a likelihood of correlation between the transaction data and transaction data of a corresponding one of the plurality of user profile nodes;
identifying, based on the probability score generated by the classifier model, a matching user profile node from among the plurality of user profile nodes based on a similarity of the transaction data; and
associating the transaction data with the matching user profile node based on the probability score, wherein associating the transaction data with the matching user profile node further comprises at least one of updating an existing edge confidence, creating a new edge, or removing an existing edge to improve the accuracy and completeness of the customer identity graph,
wherein updating the existing edge confidence includes adjusting a strength of association between the matching user profile node to one or more user profile nodes within the customer identity graph,
wherein creating the new edge includes, in response to determining that the probability score is above an upper identity edge threshold, creating a new edge between the matching user profile node and one or more user profile nodes within the customer identity graph, and
wherein removing the existing edge includes, in response to determining that a probability scores is below a lower identity edge threshold, removing the existing edge between the matching user profile node and one or more user profile nodes within the customer identity graph;
receiving a request for a customer identity of the customer, the request comprising attributes of the customer and a desired identity confidence:
in response to receiving the request, determining a user cluster from among the plurality of user clusters corresponding to the customer identity; and
identifying the user profile nodes associated with the user cluster node based on the received attributes and the desired identity confidence, wherein identifying the user profile nodes further comprises identifying one or more customer accounts associated with the user profile nodes within the determined user cluster that satisfy the desired identity confidence based on a node confidence of the user profile nodes corresponding to the one or more customer accounts; and
providing an identification of the one or more customer accounts identified by the user profile nodes within the determined user cluster.
8 . The customer identity management platform of claim 7 , wherein the transaction data further includes a card number of the third party payment card.
9 . The customer identity management platform of claim 8 , wherein the computing system is communicatively connected to the third party service and configured to submit at least a portion of the transaction data to the third party service by providing the customer name, the location of the retail enterprise at which the transaction occurred, and a deidentified representation of the card number.
10 . A non-transitory computer-readable storage medium comprising computer-executable instructions which, when executed, cause a customer identity management platform of a computing system to perform actions comprising:
obtaining transaction data associated with a third party payment card of a customer used in a transaction at a location of retail enterprise, the transaction data including transaction attributes including at least a customer name, a card identifier of the third party payment card, and the location of the retail enterprise at which the transaction occurred;
submitting at least a portion of the transaction data to a third party service after completion of a transaction represented by the transaction data, wherein the third party service is external to the retail enterprise, at least a portion of the transaction data including the customer name, the location of the retail enterprise at which the transaction occurred, and a deidentified representation of the card identifier, wherein submitting a portion of the transaction data includes deidentifying the transaction data associated with the third party payment card;
receiving, from the third party service, a plurality of potential customer contact information data sets of the customer in response to submitting at least a portion of the transaction data to the third party service, the plurality of potential customer contact information data sets including a plurality of potential customer addresses, emails, and phone numbers;
identifying, within an identity graph maintained by the identity management platform of the computing system, a plurality of user profile nodes corresponding to the plurality of potential customer contact information data sets of the customer, the customer identity graph including a plurality of user clusters, each user cluster being associated with a different customer and including one or more user profile nodes, each of the one or more user profile nodes being associated with a customer account of the customer, wherein the identity management platform further includes a random forest classifier model trained using transaction data and account to transaction correlations representing pairs of known correlated nodes or pairs of non-correlated nodes;
comparing, via the random forest classifier model, the transaction data to other transaction data associated with the plurality of profile nodes, wherein the identity management platform further includes the random forest classifier model;
generating, via the classifier model, based on the comparison, a probability score, the probability score representing a likelihood of correlation between the transaction data and transaction data of a corresponding one of the plurality of user profile nodes;
identifying, based on the correlation probability score generated by the classifier model, a matching user profile node from among the plurality of user profile nodes based on a similarity of the transaction data;
associating the transaction data with the matching user profile node based on the probability score, wherein associating the transaction data with the matching user profile node further comprises at least one of updating an existing edge confidence based at least in part on the similarity of the transaction data, creating a new edge, or removing an existing edge to improve the accuracy and completeness of the customer identity graph,
wherein updating the existing edge confidence includes adjusting a strength of association between the matching user profile node to one or more user profile nodes within the customer identity graph,
wherein creating the new edge includes, in response to determining that the probability score is above an upper identity edge threshold, creating the new edge between the matching user profile node and one or more user profile nodes within the customer identity graph, and
wherein removing the existing edge includes, in response to determining that a probability scores is below a lower identity edge threshold, removing the existing edge between the matching user profile node and one or more user profile nodes within the customer identity graph;
receiving a request for a customer identity of the customer, the request comprising attributes of the customer and a desired identity confidence;
in response to receiving the request, determining a user cluster from among the plurality of user clusters corresponding to the customer identity;
identifying the user profile nodes associated with the user cluster node based on the received attributes and the desired identity confidence, wherein identifying the user profile nodes further comprises identifying one or more customer accounts associated with the user profile nodes within the determined user cluster that satisfy the desired identity confidence based on a node confidence of the user profile nodes corresponding to the one or more customer accounts; and
providing an identification of the one or more customer accounts identified by user profile nodes within the determined user cluster.
11 . The non-transitory computer-readable storage medium of claim 10 , wherein each of the user profile nodes included in the identity graph has a node confidence, the node confidence being based in part on account attributes of the customer account associated with the corresponding user profile node, and each edge between a pair of user profile nodes has an edge confidence based in part on a similarity determination between attributes of the pair of use profile nodes.
12 . The non-transitory computer-readable storage medium of claim 11 , further comprising updating the node confidence of the matching user profile node includes upgrading the node confidence from a low node confidence derived from account attributes of the customer account corresponding to the matching user profile node to a higher node confidence based on both the account attributes and the similarity of the transaction data to the other transaction data of the one or more user profile nodes.
13 . The non-transitory computer-readable storage medium of claim 10 , wherein the node confidence of the matching user profile node is based at least in part on an observed accuracy of account attribute data associated with the third party payment card.
14 . The non-transitory computer-readable storage medium of claim 13 , wherein updating the node confidence includes increasing the node confidence in response to identifying the matching user profile node.
15 . The non-transitory computer-readable storage medium of claim 10 , wherein the instructions further cause the identity management platform of the computing system to perform:
in response to failing to identify a matching user profile node, maintaining a current confidence level of each of the plurality of user profile nodes corresponding to the plurality of potential customer addresses.
16 . The non-transitory computer-readable storage medium of claim 10 , wherein the comparison of the transaction data to other transaction data associated with the plurality of user profile nodes includes an identification of at least one of a common item preference between items represented in the transaction data and the other transaction data or a similar item purchasing pattern reflected in the transaction data and the other transaction data.