Computer systems and methods for building and analyzing data graphs
A computing platform may be configured to (i) obtain an input dataset, (ii) construct a graph from the input dataset, (iii) for a given node within the constructed graph, generate a first type of embedding vector using a first embedding technique (e.g., a shallow embedding technique) and a second type of embedding vector using a second embedding technique that differs from the first embedding technique (e.g., a deep embedding technique), and (iv) use the first and second types of embedding vectors for the given node and a data science model to render a given prediction for the given node.
1 . A computing platform comprising:
a network interface; at least one processor;
at least one non-transitory computer-readable medium; and program instructions stored on the at least one non-transitory computer-readable medium that, when executed by the at least one processor, cause the computing platform to:
provide, for access by a client device, a network-accessible interface to a toolkit of a computing device for configuring graph analysis pipelines;
receive, from the client device via the network-accessible interface, configuration data for a graph analysis pipeline, wherein the configuration data comprises:
(i) a first type of configuration data defining an input dataset for the graph analysis pipeline, wherein the first type of configuration data comprises data identifying one or more data sources from which the input dataset is to be obtained;
(ii) a second type of configuration data defining functionality to be carried out by the graph analysis pipeline for constructing at least one graph based on the input dataset, wherein the second type of configuration data comprises data identifying types of nodes to be included in the at least one graph; and
(iii) a third type of configuration data defining functionality to be carried out by the graph analysis pipeline for analyzing the constructed at least one graph, wherein the third type of configuration data comprises data identifying two or more of a deterministic analysis task, a deterministic feature extraction task, a probabilistic feature generation task, a model training task, and a model execution task;
based on the received configuration data for the graph analysis pipeline, automatically configure the graph analysis pipeline;
automatically deploy, based on a type of computing resources that are to be allocated for deployment of the graph analysis pipeline or an extent of the computing resources that are to be allocated for the deployment of the graph analysis pipeline, the graph analysis pipeline on at least one computing system; and
after deploying the graph analysis pipeline, present via the network-accessible interface, two or more of (i) a visualization of the constructed at least one graph, (ii) data produced by the functionality for analyzing the constructed at least one graph, and (iii) explainer data associated with the data produced by the functionality for analyzing the constructed at least one graph.
2 . The computing platform of claim 1 , wherein the second type of configuration data defining the functionality to be carried out by the graph analysis pipeline for constructing the at least one graph based on the input dataset further comprises configuration data identifying at least a subset of types of edges to be included in the at least one graph, a set of node attributes to be included in the at least one graph, or a set of edge attributes to be included in the at least one graph.
3 . The computing platform of claim 2 , wherein the second type of configuration data defining the functionality to be carried out by the graph analysis pipeline for constructing the at least one graph based on the input dataset further comprises configuration data identifying one or more data transformations to be applied to the input dataset.
4 . The computing platform of claim 1 , wherein the configuration data for the graph analysis pipeline is received as (i) a set of structured data, (ii) one or more configuration files containing the configuration data, (iii) an identifier of one or more configuration files containing the configuration data, or (iv) user input provided via a user interface.
5 . The computing platform of claim 1 , further comprising program instructions stored on the at least one non-transitory computer-readable medium that, when executed by the at least one processor, cause the computing platform to: based on the configuration data for the graph analysis pipeline, determine one or both of (i) a type of computing resources that are to be allocated for the deployment of the graph analysis pipeline or (ii) an extent of the computing resources that are to be allocated for the deployment of the graph analysis pipeline.
6 . The computing platform of claim 1 , wherein the configuration data for the graph analysis pipeline is received as (i) a set of structured data, (ii) one or more configuration files containing the configuration data, (iii) an identifier of one or more configuration files containing the configuration data, or (iv) user input provided via a user interface.
7 . A computer-implemented method comprising:
providing, for access by a client device, a network-accessible interface to a toolkit of a computing device for configuring graph analysis pipelines;
receiving, from the client device via the network-accessible interface, configuration data for a graph analysis pipeline, wherein the configuration data comprises:
(i) a first type of configuration data defining an input dataset for the graph analysis pipeline, wherein the first type of configuration data comprises data identifying one or more data sources from which the input dataset is to be obtained;
(ii) a second type of configuration data defining functionality to be carried out by the graph analysis pipeline for constructing at least one graph based on the input dataset, wherein the second type of configuration data comprises data identifying types of nodes to be included in the at least one graph; and
(iii) a third type of configuration data defining functionality to be carried out by the graph analysis pipeline for analyzing the constructed at least one graph, wherein the third type of configuration data comprises data identifying two or more of a deterministic analysis task, a deterministic feature extraction task, a probabilistic feature generation task, a model training task, and a model execution task;
based on the received configuration data for the graph analysis pipeline, automatically configuring the graph analysis pipeline;
automatically deploying, based on a type of computing resources that are to be allocated for deployment of the graph analysis pipeline or an extent of the computing resources that are to be allocated for the deployment of the graph analysis pipeline, the graph analysis pipeline on at least one computing system; and
after deploying the graph analysis pipeline, presenting via the network-accessible interface, two or more of (i) a visualization of the constructed the at least one graph, (ii) data produced by the functionality for analyzing the constructed at least one graph, and (iii) explainer data associated with the data produced by the functionality for analyzing the constructed at least one graph.
8 . The computer-implemented method of claim 7 , wherein:
the second type of configuration data defining the functionality to be carried out by the graph analysis pipeline for constructing the at least one graph based on the input dataset further comprises configuration data identifying at least a subset types of edges to be included in the at Least one graph, a set of node attributes to be included in the at least one graph, or a set of edge attributes to be included in the at least one graph.
9 . The computer-implemented method of claim 7 , further comprising:
based on the configuration data for the graph analysis pipeline, determining one or both of (i) a type of computing resources that are to be allocated for the deployment of the graph analysis pipeline or (ii) an extent of the computing resources that are to be allocated for the deployment of the graph analysis pipeline.
10 . The computer-implemented method of claim 7 , wherein the second type of configuration data defining the functionality to be carried out by the graph analysis pipeline for constructing the at least one graph based on the input dataset further comprises configuration data identifying one or more data transformations to be applied to the input dataset.
11 . A non-transitory computer-readable medium, wherein the non-transitory computer-readable medium is provisioned with program instructions that, when executed by at least one processor, cause a computing platform to perform a set of functions comprising:
providing, for access by a client device, a network-accessible interface to a toolkit of a computing device for configuring graph analysis pipelines;
receiving, from the client device via the network-accessible interface, configuration data for a graph analysis pipeline, wherein the configuration data comprises:
(i) a first type of configuration data defining an input dataset for the graph analysis pipeline, wherein the first type of configuration data comprises data identifying one or more data sources from which the input dataset is to be obtained;
(ii) a second type of configuration data defining functionality to be carried out by the graph analysis pipeline for constructing at least one graph based on the input dataset, wherein the second type of configuration data comprises data identifying types of nodes to be included in the at least one graph; and
(iii) a third type of configuration data defining functionality to be carried out by the graph analysis pipeline for analyzing the constructed at least one graph, wherein the third type of configuration data comprises data identifying two or more of a deterministic analysis task, a deterministic feature extraction task, a probabilistic feature generation task, a model training task, and a model execution task;
based on the received configuration data for the graph analysis pipeline, automatically configuring the graph analysis pipeline;
automatically deploying, based on a type of computing resources that are to be allocated for deployment of the graph analysis pipeline or an extent of the computing resources that are to be allocated for the deployment of the graph analysis pipeline, the graph analysis pipeline on at least one computing system; and
after deploying the graph analysis pipeline, presenting via the network-accessible interface, two or more of (i) a visualization of the constructed at least one graph, (ii) data produced by the functionality for analyzing the constructed at least one graph, and (iii) explainer data associated with the data produced by the functionality for analyzing the constructed at least one graph.