Developing object ontologies and data usage models using machine learning
Inventors
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11748395-B2
Publication Date
2023-09-05
Expiration Date
Abstract
An enterprise ontology, an application data usage model, and/or cross-application data dependencies may be developed using artificial intelligence. Using pattern recognition and/or information extraction techniques, the artificial intelligence may analyze application source code to identify common DDL or SQL statements to formulate an ontology and/or a usage model for the application. A plurality of application ontologies and/or data usage models may be used to build a semantic hub. The semantic hub may be analyzed to identify data redundancies, data use frequency, potential data quality challenges, and/or data dependencies between applications to produce a data abstraction model that allows legacy applications to communicate with one or more data stores.
Core Innovation
The invention ingests data accessed by a first application from one or more data stores and generates, based on the ingested data, a data definition language repository that catalogs database object metadata associated with the data accessed by the first application. It also accesses one or more code repositories to obtain source code of the first application.
Using natural language processing, the invention analyzes the source code of the first application to identify one or more data statements that access the one or more data stores. It identifies, using a machine learning model, a link between the one or more data statements and the database object metadata.
Based on the link, the invention generates a first application data usage model illustrating a relationship between the first application and the one or more data stores. Based on the first application data usage model and at least one second application data usage model, it generates a semantic hub and, based on the semantic hub, a data abstraction model configured to transform data exchanges between a second application and at least one data store of the one or more data stores.
Claims Coverage
The document includes three independent claims covering the same core pipeline, with five inventive features expressed through data ingestion, repository/catalog generation, NLP/ML-based source-code analysis and semantic linking, generation of application data usage models, creation of a semantic hub, and generation of a data abstraction model for transforming data exchanges.
Ingested data to DDL repository of database object metadata
ingesting data accessed by a first application from one or more data stores; generating, based on the ingested data, a data definition language repository that catalogs database object metadata associated with the data accessed by the first application.
NLP analysis of source code to find data statements accessing data stores
accessing one or more code repositories to obtain source code of the first application; analyzing, using natural language processing, the source code of the first application to identify one or more data statements that access the one or more data stores.
Machine learning link between data statements and database object metadata
identifying, using a machine learning model, a link between the one or more data statements and the database object metadata.
Data usage model generation and semantic hub synthesis
generating, based on the link between the one or more data statements and the database object metadata, a first application data usage model illustrating a relationship between the first application and the one or more data stores; generating, based on the first application data usage model and based on at least one second application data usage model, a semantic hub.
Data abstraction model to transform data exchanges
generating, based on the semantic hub, a data abstraction model configured to transform data exchanges between a second application and at least one data store of the one or more data stores.
Across the independent claims, coverage centers on generating application data usage models by linking NLP-identified data statements in application source code to a DDL repository of database object metadata, synthesizing a semantic hub from one or more usage models, and using the semantic hub to generate a data abstraction model that transforms data exchanges between an application and data stores.
Stated Advantages
Documented Applications
No documented applications found
Interested in licensing this patent?