Applied artificial intelligence technology for natural language generation using a graph data structure in combination with chooser code, structurer code, and realizer code
Natural language generation technology is disclosed that applies artificial intelligence to structured data to determine content for expression in natural language narratives that describe the structured data. A graph data structure is employed, where the graph data structure comprises a plurality of nodes. Each of a plurality of the nodes (1) represents a corresponding intent so that a plurality of different nodes represent different corresponding intents and (2) is associated with one or more links to one or more of the nodes to define relationships among the intents. A processor executes chooser code, structurer code, and realizer code to generate the natural language narratives. The chooser code determines content for expression in a natural language narrative based on a traversal of the graph data structure based on defined criteria and a plurality of the links, wherein the traversal operates to choose a plurality of the nodes for evaluation with respect to the structured data to derive content for expression in that natural language narrative.
1 . A natural language generation (NLG) system that applies artificial intelligence to structured data to determine content to be expressed in a natural language narrative that describes the structured data, the system comprising:
a memory configured to store a graph data structure in memory via a processor, the graph data structure comprising a plurality of nodes, each of a plurality of the nodes (1) representing a corresponding intent so that a plurality of different nodes represent different corresponding intents, (2) representing a corresponding computation capable of being performed on the structured data, and (3) being associated with one or more links to one or more of the nodes to define relationships among the intents; and
a processor configured to (1) execute chooser code to determine content for expression in the natural language narrative based on a traversal of the graph data structure based on defined criteria and a plurality of the links, the traversal operating to choose a subset of the plurality of nodes for evaluation with respect to the structured data to derive content for expression in the natural language narrative, (2) perform one or more computations on the structured data to determine one or more numerical outcome values based on the corresponding computations associated with the subset of the plurality of nodes; (3) execute structurer code to organize the determined content and the numerical outcome values into an outline for the natural language narrative, the outline including a plurality of sections corresponding to portions of the natural language narrative, and
(4) execute realizer code to generate the natural language narrative that describes the structured data based on the outline, the generated natural language narrative expressing the determined content including the numerical outcome values in accordance with the outline.
2 . The system of claim 1 wherein the chooser code generates a story graph that identifies the determined content.
3 . The system of claim 2 wherein the structurer code translates the story graph into the outline based on a plurality of rules.
4 . The system of claim 1 wherein the chooser code comprises a plurality of choosers, and wherein the processor is further configured to select which of the choosers is to operate on the knowledge graph with respect to the structured data to generate the natural language narrative.
5 . The system of claim 4 wherein the different choosers are configured with different operating rules and/or parameters so that different choosers will traverse the knowledge graph differently with respect to determining content for inclusion in a natural language narrative.
6 . The system of claim 4 wherein the processor operates to select different choosers to determine content for inclusion in different natural language narratives.
7 . The system of claim 1 wherein the structurer code comprises a plurality of structurers, and wherein the processor is further configured to select which of the structurers is to operate on the determined content to generate the outline.
8 . The system of claim 7 wherein the different structurers are configured with different operating rules and/or parameters so that different structurers will organize the determined content differently into different outlines.
9 . The system of claim 7 wherein the processor operates to select different structurers to generate different outlines for different natural language narratives.
10 . The system of claim 1 wherein the chooser code and/or the structurer code comprises plug-in code.
11 . The system of claim 1 wherein the graph data structure is adjustable to add one or more additional nodes to the graph data structure.
12 . The system of claim 1 wherein the graph data structure is adjustable modify one or more of the nodes.
13 . The system of claim 1 wherein the graph data structure is parameterized based on the structured data.
14 . The system of claim 1 wherein the graph data structure comprises an authoring graph.
15 . The system of claim 14 wherein the authoring graph is parameterized based on the structured data to define a knowledge graph, wherein the chooser code operates on the knowledge graph to determine content for expression in the natural language narrative.
16 . The system of claim 1 wherein the processor is configured to generate a plurality of natural language narratives based on the graph data structure in an interactive mode based on conversational inputs from users.
17 . The system of claim 1 wherein the processor comprises a plurality of processors.
18 . A natural language generation (NLG) method that applies artificial intelligence to structured data to determine content to be expressed in a natural language narrative that describes the structured data, the method comprising:
accessing a graph data structure in memory via a processor, the graph data structure comprising a plurality of nodes, each of a plurality of the nodes (1) representing a corresponding intent so that a plurality of different nodes represent different corresponding intents, (2) representing a corresponding computation capable of being performed on the structured data, and (3) being associated with one or more links to one or more of the nodes to define relationships among the intents;
executing chooser code via the processor to determine content for expression in the natural language narrative based on a traversal of the graph data structure based on defined criteria and a plurality of the links, the traversal operating to choose a subset of the plurality of nodes for evaluation with respect to the structured data to derive content for expression in the natural language narrative;
performing one or more computations on the structured data to determine one or more numerical outcome values based on the corresponding computations associated with the subset of the plurality of nodes;
executing structurer code via the processor to organize the determined content and the numerical outcome values into an outline for the natural language narrative, the outline including a plurality of sections corresponding to portions of the natural language narrative; and
executingrealizer code via the processor to generate the natural language narrative that describes the structured data based on the outline, the generated natural language narrative expressing the determined content including the numerical outcome values in accordance with the outline.
19 . The method of claim 18 wherein the processor comprises a plurality of processors, wherein different processors execute the chooser code, the structurer code, and/or the realizer code.
20 . One or more non-transitory computer readable media having instructions stored thereon for performing a natural language generation (NLG) method that applies artificial intelligence to structured data to determine content to be expressed in a natural language narrative that describes the structured data, the method comprising:
accessing a graph data structure in memory via a processor, the graph data structure comprising a plurality of nodes, each of a plurality of the nodes (1) representing a corresponding intent so that a plurality of different nodes represent different corresponding intents, (2) representing a corresponding computation capable of being performed on the structured data, and (3) being associated with one or more links to one or more of the nodes to define relationships among the intents;
executing chooser code via the processor to determine content for expression in the natural language narrative based on a traversal of the graph data structure based on defined criteria and a plurality of the links, the traversal operating to choose a subset of the plurality of nodes for evaluation with respect to the structured data to derive content for expression in the natural language narrative;
performing one or more computations on the structured data to determine one or more numerical outcome values based on the corresponding computations associated with the subset of the plurality of nodes;
executing structurer code via the processor to organize the determined content and the numerical outcome values into an outline for the natural language narrative, the outline including a plurality of sections corresponding to portions of the natural language narrative; and
executing realizer code via the processor to generate the natural language narrative that describes the structured data based on the outline, the generated natural language narrative expressing the determined content including the numerical outcome values in accordance with the outline.