Using email history to estimate creditworthiness for applicants having insufficient credit history

Patent metadata

Field	Value
Publication number	US-11915313-B2
Application number	US-202117445136-A
Country	US
Kind code	B2
Filing date	Aug 16, 2021
Priority date	Aug 16, 2021
Publication date	Feb 27, 2024
Grant date	Feb 27, 2024

Abstract

In some implementations, a credit decision platform may receive a credit request from an applicant and obtain domestic historical data associated with the applicant from a credit bureau device. The credit decision platform may obtain access to an email account associated with the applicant based on determining that the domestic historical data associated with the applicant is insufficient to process the credit request. The credit decision platform may identify, using one or more machine learning models, a set of email messages included in the email account that are relevant to the credit request and may analyze content included in the set of email messages to generate non-domestic historical data associated with the applicant. The credit decision platform may generate a decision on the credit request based on an estimated creditworthiness of the applicant, which may be determined based on the non-domestic historical data.

First claim

What is claimed is:

1. A system for estimating creditworthiness, the system comprising:
one or more memories; and
one or more processors, communicatively coupled to the one or more memories, configured to:
obtain access to an email account associated with an applicant based on the applicant having insufficient domestic historical data for a credit request of the applicant to be processed;
train a plurality of machine learning models to distinguish authentic email accounts from synthetic email accounts,
wherein a first machine learning model, of the plurality of machine learning models, is configured to use an input that includes a set of observations from historical data and generate an output that indicates whether the email account is authentic or synthetic using a target variable,
wherein the set of observations include a feature set which is extracted by a second machine learning model, of the plurality of machine learning models, from the historical data by performing natural language processing, and
wherein the plurality of machine learning models are trained to recognize patterns in the feature set that lead to a value of the target variable that determines whether the email is authentic or synthetic, and
wherein the patterns in the feature set include one or more of:
 a pattern of received emails,
 a pattern of email header information, a pattern of how email messages marked as spam are handled, or
 a pattern relating to reading behavior;

identify, using the plurality of trained machine learning models, a set of email messages included in the email account that are relevant to the credit request from an email server,
wherein the set of email messages relevant to the credit request are identified, using the plurality of trained machine learning models, based on a determination that the email account is authentic;

analyze content included in the set of email messages using the plurality of trained machine learning models and the natural language processing to generate non-domestic historical data associated with the applicant;
determine a set of metrics that relate to an estimated creditworthiness of the applicant based on the non-domestic historical data associated with the applicant;
generate a decision that the credit request is granted based on the target variable indicating that emails in the set of email messages are authentic and the set of metrics that relate to the estimated creditworthiness of the applicant; and
provide information to allow the applicant to open a credit account based on the decision.

2. The system of claim 1 , wherein the plurality of machine learning models are configured to generate an output that indicates whether the email account is authentic or synthetic based on one or more metrics that relate to patterns associated with email messages included in the email account or patterns in user behaviors that occur during interactions with the email account.

3. The system of claim 1 , wherein the credit request is a first credit request, and
wherein the plurality of machine learning models are configured to generate one or more outputs for a second credit request that indicate whether individual email messages in the email account are authentic or synthetic based on content or metadata associated with the individual email messages.

4. The system of claim 3 , wherein the second credit request is denied based on the one or more outputs indicating that a threshold number of the individual email messages are synthetic.

5. The system of claim 1 , wherein the content included in the set of email messages is analyzed, using the natural language processing to identify words or phrases that relate to financial activities, to identify values associated with the financial activities.

6. The system of claim 1 , wherein the one or more processors are further configured to:
determine that at least a portion of the content included in the set of email messages is associated with one or more languages other than a language used in the natural language processing; and
cause the portion of the content to be reformatted or translated into the language used in the natural language processing.

7. The system of claim 1 , wherein the applicant is a first applicant, the credit request is a joint credit request with a second applicant having insufficient domestic historical data to process the joint credit request, and the one or more processors are further configured to:
obtain access to an email account associated with the second applicant;
identify, using the plurality of machine learning models, a set of email messages included in the email account associated with the second applicant that are relevant to the credit request;
analyze content included in the set of email messages associated with the second applicant to generate non-domestic historical data associated with the second applicant; and
generate the decision on the joint credit request based on a cross-referencing of the non-domestic historical data associated with the first applicant and the non-domestic historical data associated with the second applicant.

8. The system of claim 1 , wherein the one or more processors, when obtaining access to the email account, are configured to:
receive a token that grants access to the email account associated with the applicant; and
provide the token to an email server associated with the email account to obtain access to the email account associated with the applicant.

9. The system of claim 8 , wherein the one or more processors are further configured to:
provide, to the email server, a request to obtain email messages included in the email account that satisfy one or more conditions related to time, content, status, or type; and
download, from the email server, email messages included in the email account that satisfy the one or more conditions.

10. The system of claim 1 , wherein the one or more processors, when generating the decision on the credit request, are further configured to:
map the estimated creditworthiness of the applicant to a credit score;
obtain, based on one or more of input from the applicant or analysis of content included in the set of email messages, information related to income or assets associated with the applicant; and
generate the decision on the credit request based on the credit score and the information related to the income or assets associated with the applicant.

11. A method for estimating creditworthiness, comprising:
receiving, by a credit decision platform, a credit request from an applicant;
obtaining, by the credit decision platform, domestic historical data associated with the applicant from one or more credit bureau devices;
obtaining, by the credit decision platform, access to an email account associated with the applicant based on determining that the domestic historical data associated with the applicant is insufficient to process the credit request;
training, by the credit decision platform, a machine learning model to distinguish authentic email accounts from synthetic email accounts,
wherein the machine learning model is configured to use an input that includes a set of observations from historical data and generate an output that indicates whether the email account is authentic or synthetic using a target variable,
wherein the set of observations include a feature set which is extracted by another machine learning model from the historical data by performing natural language processing,
wherein the machine learning model is trained to recognize patterns in the feature set that lead to a value of the target variable that determines whether the email is authentic or synthetic, and
wherein t

Assignees

• Capital One Services Llc

Inventors

• Cheng Lin Ni Lisa
• Edwards Joshua
• Atkins Phoebe
• Miracolo Max
• Vargas Cruz
• Mcclanahan Brian
• Lin Alexander
• Buell Louis
• Mossoba Michael

Classifications

• G06Q40/03 Primary

  Credit; Loans; Processing thereof · CPC title

• G06F40/289

  Phrasal analysis, e.g. finite state techniques or chunking · CPC title

• G06F40/58

  Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation · CPC title

• G06N20/00

  Machine learning · CPC title

• G06F40/263

  Language identification · CPC title