Back to results
US 12,596,604 B2Granted
Systems and methods for data management
AssigneeCapital One Services, LLC
InventorManav Jain
Life of this patent
Estimating the term…
Abstract
A data management method may include receiving a use case, and receiving a configuration specific to the use case associated with an item type having a plurality of attributes. The configuration may define a target data store, a subset of the plurality of attributes associated with the use case, and a subset of optimal data sources from a plurality of available data sources each configured to provide one or more attributes within the subset of the plurality of attributes. The data management method may further include, based on the configuration, accessing, from the subset of optimal data sources, values corresponding to the subset of the plurality of attributes, and storing the values in the target data store.
Claims20
A method for data management, the method comprising:
- receiving a data management use case;
- determining, using a trained machine learning model, a configuration based on the use case, wherein the configuration is associated with an item type having a plurality of attributes, and the configuration defines:
- a target data store;
- a subset of the plurality of attributes associated with the use case;
- and
- a subset of optimal data sources from a plurality of available data sources each configured to provide one or more attributes within the subset of the plurality of attributes;
- based on the configuration, accessing data from the subset of optimal data sources, the data including values corresponding to the subset of the plurality of attributes;
- storing the values in the target data store;
- performing an independent control process based on the data accessed from the subset of optimal data sources, the configuration, and the values stored in the target data store;
- detecting, based on the independent control process, an error associated with one or more of a presence, a format, or a substantive accuracy of a first value corresponding to a first attribute, from the values corresponding to the subset of the plurality of attributes, the error indicating a first optimal data source of the subset of optimal data sources from which the first value was obtained is a non-optimal data source for providing the first attribute;
- and
- receiving, as feedback from the independent control process, a second optimal data source for obtaining the first attribute to remedy the error and a reason for the error associated with the first optimal data source, wherein a second value corresponding to the first attribute is accessed from the second optimal data source, the first value is replaced with the second value in the target data store, and the trained machine learning model is updated based on the feedback to learn a relationship between the reason for the error and the second optimal data source for use in defining optimal data sources in future configuration determinations received when the independent control process detects an error.
9 more elements
A method for data management, the method comprising:
- receiving a data management use case;
- receiving a configuration specific to the use case associated with an item type having a plurality of attributes, the configuration defining:
- a first target data store and a second target data store;
- and
- for each of the first target data store and the second target data store:
- a subset of the plurality of attributes associated with the use case;
- and
- a subset of optimal data sources from a plurality of available data sources each configured to provide one or more attributes within the subset of the plurality of attributes;
- based on the configuration, accessing data values from the respective subset of optimal data sources for each of the first target data store and the second target data store, the data values corresponding to the subset of the plurality of attributes;
- respectively storing the data values in each of the first target data store and the second target data store;
- validating the configuration based on the data values;
- based on the validating, identifying an error associated with one or more of the data values accessed from a first optimal data source of the respective subset of optimal data sources;
- in response to determining a percentage of data values accessed from the first optimal data source for which errors are identified is above a threshold value, excluding the first optimal data source from the plurality of available data sources in future configuration determinations;
- and
- enriching a system of record based on the first target data store and the second target data store.
12 more elements
A method for data management, the method comprising:
- receiving a plurality of use cases;
- for each use case of the plurality of use cases:
- receiving a configuration specific to the use case associated with an item type having a plurality of attributes, the configuration defining:
- a target data store;
- a subset of the plurality of attributes associated with the use case;
- and
- a subset of optimal data sources from a plurality of available data sources each configured to provide one or more attributes within the subset of the plurality of attributes;
- based on the configuration, accessing, from the subset of optimal data sources, values corresponding to the subset of the plurality of attributes;
- storing the values in the target data store;
- and
- performing an independent control process to detect any errors associated with the values stored in the target data store, wherein detecting an error indicates at least one of the subset of optimal data sources is a non-optimal data source for providing a respective attribute within the subset of the plurality of attributes;
- determining, based on the independent control process performed for each use case of the plurality of use cases, an amount of data accessed from an available data source of the plurality of available data sources resulting in detected errors is above a threshold;
- and
- based on the determination, excluding the available data source from the plurality of available data sources in future configuration determinations.
11 more elements
Dependencies are inferred from the claim text.