Method and system for using machine learning techniques to make highly relevant and de-duplicated offer recommendations

A system for using machine learning techniques to make highly relevant and de-duplicated offer recommendations comprising:

• one or more memory systems, the one or more memory systems including:
• offer data representing one or more historical or current offers, the offer data being obtained, at least in part, from, or through, one or more providers of products or services;
• model training data, the model training data being derived from raw user data including one or more of, profile data, behavior data, preference data, transaction data, and credit history data associated with two or more users, the model training data including one or more of, profile model training data, behavior model training data, preference model training data, transaction model training data, and credit history model training data associated with two or more users;
• a user feature engineering module, the user feature engineering module including a personal attribute extraction model, the personal attribute extraction model of the user feature engineering module having been trained using one or more of, the profile model training data, the behavior model training data, and the preference model training data, the personal attribute extraction model of the user feature engineering module generating personal attribute data associated with two or more users;
• a product management module, the product management module including an historical offer extraction engine and a financial attribute extraction model, the financial attribute extraction model of the product management module having been trained using the one or more of the transaction model training data and the credit history model, the financial attribute extraction model of the product management module generating financial attribute data associated with two or more users;
• an offer management module, the offer management module including an offer attribute extraction engine, the offer attribute extraction engine receiving the offer data as input data and generating offer attribute data as output data;
• a feature engineering module, the feature engineering module including a user/offer attribute matching model, the user/offer attribute matching model having been trained using one or more of, the personal attribute data from the personal attribute extraction model of the user feature engineering module, the financial attribute data from the financial attribute extraction model of the product management module, and the offer attribute data from the offer attribute extraction engine of the offer management module, the user/offer attribute matching model generating user/offer attribute matching data associated with two or more users;
• an offer recommendation module, the offer recommendation module including an offer de-duplication engine and an offer recommendation model, the offer recommendation model having been trained using one or more of the user/offer attribute matching data from the feature engineering module, the personal attribute data from the personal attribute extraction model of the user feature engineering module, the financial attribute data from the financial attribute extraction model of the product management module, and the offer attribute data from the offer attribute extraction engine of the offer management module, the offer recommendation model of the offer recommendation module generating offer recommendation data;
• user's data associated with a given user obtained from one or more sources of user's data, the user's data including one or more of user's profile data, user's behavior data, user's preference data, user's transaction data, and user's credit history data;
• and
• one or more processors, the one or more processors executing instructions for performing a runtime execution of a process for using machine learning techniques to make highly relevant and de-duplicated offer recommendations, the process for using machine learning techniques to make highly relevant and de-duplicated offer recommendations including:
• providing one or more of the user's profile data, the user's behavior data, and the user's preference data, to the user feature engineering module;
• executing the user feature engineering module to generate user's personal attribute data;
• providing one or more of the user's transaction data and the user's credit history data to the product management module;
• executing the product management module to generate user's historical offer data and user's financial attribute data;
• providing one or more of the user's personal attribute data, the user's financial attribute data, and the offer attribute data, to the feature engineering module;
• executing the feature engineering module to generate user's/offer attribute matching data;
• providing one of more of the user's/offer attribute matching data, the offer data, and the user's historical offer data, to the de-duplication engine of the offer recommendation module;
• executing the de-duplication engine of the offer recommendation module to generate de-duplicated offer data;
• providing one or more of the de-duplicated offer data, the user's profile data, the user's behavior data, the user's preference data, the user's transaction data, and the user's credit history data, to the offer recommendation module;
• executing the offer recommendation module to generate offer recommendation data;
• and
• providing at least part of the offer recommendation data to a user computing system associated with, or accessible by, the given user.

A computing system implemented method for using machine learning techniques to make highly relevant and de-duplicated offer recommendations comprising:

• obtaining raw user data from one or more sources of raw user data, the raw user data including one or more of, profile data, behavior data, preference data, transaction data, and credit history data associated with two or more users;
• processing the raw user data to transform the raw user data into model training data, the model training data including one or more of, profile model training data, behavior model training data, preference model training data, transaction model training data, and credit history model training data;
• training, using the model training data, one or more user attribute extraction models;
• training, using the model training data, one or more user/offer matching models;
• training, using the model training data, one or more offer recommendation models;
• obtaining offer data representing one or more historical or current offers;
• extracting, using one or more offer attribute extraction engines, offer attribute data from the offer data;
• obtaining user's data associated with a given user from one or more sources of user's data, the user's data including one or more of user's profile data, user's behavior data, user's preference data, user's transaction data, and user's credit history data;
• generating, using the user's transaction data and user's credit history data, user's historical offer data;
• providing one or more of the user's profile data, the user's behavior data, the user's preference data, the user's transaction data, and the user's credit history data as input data to the trained one or more user attribute extraction models, the one or more trained user attribute extraction models generating user's attribute data as output data;
• providing one or more of the user's attribute data and the offer attribute data as input data to the trained one or more user/offer matching models, the trained one or more user/offer matching models generating user's/offer attribute matching data as output data;
• providing one of more of the user's/offer attribute matching data, the offer data, and the user's historical offer data as input data to a de-duplication engine, the de-duplication engine generating de-duplicated offer data as output data;
• providing one or more of the de-duplicated offer data, the user's profile data, the user's behavior data, the user's preference data, the user's transaction data, and the user's credit history data, as input data to the trained one or more offer recommendation models, the trained one or more offer recommendation models generating offer recommendation data as output data;
• and
• providing at least part of the offer recommendation data to a user computing system associated with, or accessible by, the given user.