US 2025/0165774 A1Published application

System and methods for data interconnections in a data ecosystem

AssigneeTruist Bank

InventorsScott Howard Magoon, Tufail Ahmed Khan, Di Hu

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method is disclosed for creating a representation of interconnections between datasets using machine learning. The system receives datasets, each having traits, stores the datasets into a catalog, and trains, via an iterative training and testing loop, a ML program utilizing a neural network to generate a trained predictive model, a training dataset utilized during the training of the ML program. The training includes inserting a target variable value and iteratively predicting the target variable via the iterative training and testing loop. The system deploys the model and predicts: (1) a common trait for a first and second dataset; (2) a representation of a first interconnection between the first and second dataset; and (3) a common trait for the second dataset and a third dataset. The system generates a representation of the second interconnection, comprising a second value, displaying a governance graph depicting the first interconnection and the second interconnection.

Claims20

Dependent claims

A computer-implemented method for creating a representation of interconnections between data sets using a machine learning model, the method comprising the steps of:

receiving a plurality of data sets from a plurality of sources using a computer, the data sets including a plurality of traits;
storing the plurality of data sets into a data catalog;
training, via an iterative training and testing loop, a machine learning program utilizing at least one neural network to generate a trained predictive model, a training data set utilized during the training of the machine learning program comprising the personal data set of each user, the training comprising:
inserting a target variable value into the iterative training and testing loop;
and
iteratively predicting the target variable via the iterative training and testing loop, wherein iterative predictions of the target variable comprise modifying weights and calculations applied to the training data set during subsequent prediction iterations in order to improve predictability of the target variable;
deploying the trained predictive model;
predicting, by the predictive model, at least one common trait for a first data set and a second data set from the plurality of data sets;
predicting, by the predictive model, a representation of a first interconnection between the first data set and the second data set based on the at least one common trait, wherein the representation of the first interconnection comprises a first value;
predicting, by the predictive model, at least one common trait for the second data set and a third data set from the plurality of data sets;
generating a representation of a second interconnection between the second data set and the third data set based on the at least one common trait, wherein the representation of the second interconnection comprises a second value;
and
displaying, via a graphical user interface, a governance graph comprising the first interconnection and the second interconnection.

10 more elements

A computer-implemented method for predicting data usefulness between data sets using a machine learning model, the method comprising the steps of:

receiving a plurality of data sets from a plurality of sources using a computer, the data sets including a plurality of traits;
storing the plurality of data sets into a data catalog;
training, via an iterative training and testing loop, a machine learning program utilizing at least one neural network to generate a trained predictive model, a training data set utilized during the training of the machine learning program comprising the data regulatory regulations, the training comprising:
inserting a target variable value into the iterative training and testing loop;
and
iteratively predicting the target variable via the iterative training and testing loop, wherein iterative predictions of the target variable comprise modifying weights and calculations applied to the training data set during subsequent prediction iterations in order to improve predictability of the target variable;
deploying the trained predictive model;
predicting, by the predictive model, at least one data set having a low demand for access;
predicting, by the predictive model, at least one data set having a high demand for access;
generating a representation of the low demand data set and the high demand data set, wherein the representation of the high demand data set is depicted larger than the low demand data set;
and
displaying, via a graphical user interface, a governance graph comprising the low demand data set and the high demand data set.

9 more elements

Dependencies are inferred from the claim text.