Systems and methods for knowledge graph-enabled cross-domain analytics using machine learning

1. A method comprising:

collating by one or more processors, data from a plurality of data sources into a first data repository;

identifying by the one or more processors, using a machine-learning model, class clusters and relationship clusters of the collated data in the first data repository, wherein the class clusters are identified based on one or more properties associated with the collated data, and the class clusters represent relationships of data and the relationship clusters represent relationships between class clusters;

generating by the one or more processors, a domain-specific semantic model as a graph-structured data model based on the identified class clusters and relationship clusters;

generating by the one or more processors, a data object model using the domain-specific semantic model and the collated data in the first data repository;

creating by the one or more processors, a first domain-specific knowledge graph by associating the data object model with the domain-specific semantic model;

creating by the one or more processors, a cross-domain analytics knowledge graph for deriving insights involving cross-domain analytics by merging the first domain-specific knowledge graph with a second domain-specific knowledge graph created from a second data repository, wherein said creating the cross-domain analytics knowledge graph comprises forming a relationship between at least one class of the first domain-specific knowledge graph and at least one class of the second domain-specific knowledge graph;

validating by the one or more processors, the identified class clusters and relationship clusters based on one or more inputs from a user received from a user interface, wherein validating the identified class clusters and relationship clusters comprises at least one of: modifying a name of a cluster by the machine-learning model based on a property of the cluster and modifying a relationship name of a relationship cluster by the machine-learning model based on a relationship between at least two clusters;

deriving by the one or more processors, one or more actionable insights corresponding to performance of one or more of an asset or a process in a facility based on utilization of the cross-domain analytics knowledge graph; and

displaying by the one or more processors, at least one actionable insight based on the one or more actionable derived insights for the cross-domain analytics knowledge graph.

2. The method of claim 1 , wherein the identifying the class clusters and relationship clusters includes using a trained neural network of class clusters and relationship clusters as the machine-learning model.

3. The method of claim 1 , wherein the generating the domain-specific semantic model includes generating the domain-specific semantic model as a Web Ontology Language (.owl) file based on the identified class clusters and relationship clusters.

4. The method of claim 1 , wherein the generating the data object model includes generating the data object model as a Resource Description Framework (.rdf) file based on the domain-specific semantic model and the collated data in the first data repository.

5. The method of claim 1 , wherein the creating the first domain-specific knowledge graph includes creating the first domain-specific knowledge graph as a graph-structured data model based on the data object model and the domain-specific semantic model.

6. The method of claim 1 , wherein the deriving insights includes:

querying the cross-domain analytics knowledge graph using data from the first data repository and data from the second data repository.

7. The method of claim 6 , wherein the deriving insights further includes:

displaying a response based on the querying the cross-domain analytics knowledge graph.

8. The method of claim 1 , further comprising:

providing the cross-domain analytics knowledge graph in an extensible object model in an Internet-of-Things platform.

9. A system comprising:

a memory storing instructions; and

one or more processors to execute the stored instructions to perform operations including:

collating data from a plurality of data sources into a first data repository;

identifying, using a machine-learning model, class clusters and relationship clusters of the collated data in the first data repository, wherein the class clusters are identified based on one or more properties associated with the collated data, and the class clusters represent relationships of data and the relationship clusters represent relationships between class clusters;

generating a domain-specific semantic model as a graph-structured data model based on the identified class clusters and relationship clusters;

generating a data object model using the domain-specific semantic model and the collated data in the first data repository;

creating a first domain-specific knowledge graph by associating the data object model with the domain-specific semantic model;

creating a cross-domain analytics knowledge graph for deriving insights involving cross-domain analytics by merging the first domain-specific knowledge graph with a second domain-specific knowledge graph created from a second data repository, wherein said creating the cross-domain analytics knowledge graph includes forming a relationship between at least one class of the first domain-specific knowledge graph and at least one class of the second domain-specific knowledge graph;

validating the identified class clusters and relationship clusters based on one or more inputs from a user received from a user interface, wherein validating the identified class clusters and relationship clusters comprises at least one of: modifying a name of a cluster by the machine-learning model based on a property of the cluster and modifying a relationship name of a relationship cluster by the machine-learning model based on a relationship between at least two clusters;

deriving one or more actionable insights corresponding to performance of one or more of an asset or a process in a facility based on utilization of the cross-domain analytics knowledge graph; and

displaying at least one actionable insight based on the one or more actionable derived insights for the cross-domain analytics knowledge graph.

10. The system of claim 9 , wherein the identifying the class clusters and relationship clusters includes using a trained neural network of class clusters and relationship clusters as the machine-learning model.

11. The system of claim 9 , wherein the generating the domain-specific semantic model includes generating the domain-specific semantic model as a Web Ontology Language (.owl) file based on the identified class clusters and relationship clusters.

12. The system of claim 9 , wherein the generating the data object model includes generating the data object model as a Resource Description Framework (.rdf) file based on the domain-specific semantic model and the collated data in the first data repository.

13. The system of claim 9 , wherein the creating the first domain-specific knowledge graph includes creating the first domain-specific knowledge graph as a graph-structured data model based on the data object model and the domain-specific semantic model.

14. The system of claim 9 , wherein the deriving insights includes:

querying the cross-domain analytics knowledge graph using data from the first data repository and data from the second data repository, and

displaying a response based on the querying the cross-domain analytics knowledge graph.

15. The system of claim 14 , further comprising:

providing the cross-domain analytics knowledge graph in an extensible object model in an Internet-of-Things platform.

16. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

collating data from a plurality of data sources into a first data repository;

identifying, using a machine-learning model, class clusters and relationship clusters of the collated data in the first data repository, wherein the class clusters are identified based on one or more properties associated with the collated data, and the class clusters represent relationships of data and the relationship clusters represent relationships between class clusters;

generating a domain-specific semantic model as a graph-structured data model based on the identified class clusters and relationship clusters;

generating a data object model using the domain-specific semantic model and the collated data in the first data repository;

creating a first domain-specific knowledge graph by associating the data object model with the domain-specific semantic model;

creating a cross-domain analytics knowledge graph for deriving insights involving cross-domain analytics by merging the first domain-specific knowledge graph with a second domain-specific knowledge graph created from a second data repository, wherein said creating the cross-domain analytics knowledge graph includes forming a relationship between at least one class of the first domain-specific knowledge graph and at least one class of the second domain-specific knowledge graph;

validating the identified class clusters and relationship clusters based on one or more inputs from a user received from a user interface, wherein validating the identified class clusters and relationship clusters comprises at least one of: modifying a name of a cluster by the machine-learning model based on a property of the cluster and modifying a relationship name of a relationship cluster by the machine-learning model based on a relationship between at least two clusters;

deriving one or more actionable insights corresponding to performance of one or more of an asset or a process in a facility based on utilization of the cross-domain analytics knowledge graph; and

displaying at least one actionable insight based on the one or more actionable derived insights for the cross-domain analytics knowledge graph.

17. The method of claim 1 , wherein said deriving insights further comprising identifying one or more faults for the derived insights, and wherein the one or more faults are identified by comparing current and predicted performances of the derived insights.

18. The method of claim 17 , further comprising;

providing one or more recommendations for the identified one or more faults, and wherein the one or more recommendations are prioritized based on at least one impact of the identified one or more faults.