System and method for using data and angles to automatically generate a narrative story

Patent metadata

Field	Value
Publication number	US-9990337-B2
Application number	US-201715664414-A
Country	US
Kind code	B2
Filing date	Jul 31, 2017
Priority date	May 13, 2010
Publication date	Jun 5, 2018
Grant date	Jun 5, 2018

Abstract

A system and method for automatically generating a narrative story receives data and information pertaining to a domain event. The received data and information and/or one or more derived features are then used to identify a plurality of angles for the narrative story. The plurality of angles is then filtered, for example through use of parameters that specify a focus for the narrative story, length of the narrative story, etc. Points associated with the filtered plurality of angles are then assembled and the narrative story is rendered using the filtered plurality of angles and the assembled points.

First claim

What is claimed is:

1. An apparatus for automatically generating a narrative story about a structured data set, the narrative story comprising text arranged in a narrative format, the apparatus comprising:
a memory in which a plurality of angle data structures are stored, each of a plurality of the angle data structures being representative of a characterization of data and having an associated applicability condition, the associated applicability condition for testing against data within or derived from the structured data set to support a determination as to whether the associated angle is an accurate characterization of the structured data set; and
a processor in cooperation with the memory, the processor configured to:
compute a plurality of derived features from the structured data set;
test data against a plurality of the applicability conditions for a plurality of the angle data structures in the memory, wherein the tested data comprises at least one of (1) data within the structured data set and (2) data within the computed derived features;
in response to the tests, (1) determine that a plurality of the angle data structures are candidate angle data structures that accurately characterize the structured data set, and (2) filter the candidate angle data structures to select at least one of the candidate angle data structures; and
automatically render the narrative story text for display using the selected at least one candidate angle data structure and at least one of (1) data within the structured data set and (2) data within the computed derived features such that the rendered narrative story text describes the structured data set in accordance with the at least one characterization represented by the selected at least one candidate angle data structure.

2. The apparatus of claim 1 wherein each of a plurality of the candidate angle data structures is further associated with an importance value, and wherein the processor is further configured to filter the candidate angle data structures based at least in part on the importance values associated with the candidate angle data structures.

3. The apparatus of claim 2 wherein the importance values comprise a plurality of fixed values associated with a plurality of the angle data structures.

4. The apparatus of claim 2 wherein the processor is further configured to compute a plurality of the importance values associated with a plurality of the angle data structures based on at least one of (1) data within the structured data set and (2) data within the computed derived features.

5. The apparatus of claim 1 wherein the angle data structures do not themselves comprise any specific text for the narrative story.

6. The apparatus of claim 1 wherein the selected at least one candidate angle data structure comprises a plurality of selected candidate angle data structures; and
wherein the processor is further configured to automatically render the narrative story text for display using the selected candidate angle data structures and at least one of (1) data within the structured data set and (2) data within the computed derived features such that the rendered narrative story text describes the structured data set in accordance with the characterizations represented by the selected candidate angle data structures.

7. The apparatus of claim 6 wherein the processor is further configured to:
prioritize the plurality of selected candidate angle data structures; and
automatically render the narrative story text for display such that the generated text describes the structured data set in accordance with the prioritization of the selected candidate angle data structures.

8. The apparatus of claim 7 wherein each of a plurality of the selected candidate angle data structures is further associated with an importance value, and wherein the processor is further configured to prioritize the plurality of selected candidate angle data structures based at least in part on their associated importance values.

9. The apparatus of claim 6 wherein the memory is further configured to store a plurality of phrasal generation routines, wherein each of a plurality of the angle data structures is associated with a phrasal generation routine from among the plurality of phrasal generation routines; and
wherein the processor is further configured to render the narrative story text for display based on an invocation of the phrasal generation routine associated with the selected candidate angle data structures.

10. The apparatus of claim 9 wherein each of a plurality of the angle data structures comprise a plurality of points that support the characterization represented by that angle data structure; and
wherein the processor is further configured to (1) select and assemble a plurality of elements of the structured data set to create an element list data structure that connects the selected elements to the points of the selected candidate angle data structures, and (2) render the narrative story text for display based on an application of the selected elements from the element list data structure to the invoked phrasal generation routine to thereby output and assemble a plurality of phrases in a natural language that describe the structured data set based on the points that support the characterizations represented by the selected candidate angle data structures.

11. The apparatus of claim 10 wherein each of a plurality of the angle data structures is associated with an importance value; and
wherein the processor is further configured to (1) determine an order for an expression of the characterizations represented by the selected candidate angle data structures within the narrative story based on the importance values associated with the selected candidate angle data structures, (2) perform the element selection for each of the selected candidate angle data structures to generate a plurality of the element list data structures, (3) generate an angle list data structure that represents the ordered angle data structures, (4) assemble the elements of the element list data structures based on the angle list data structure to generate an ordered element list data structure, and (5) render the narrative story text for display based on the ordered element list data structure and an application of the elements from the ordered element list data structure to the invoked phrasal generation routine to thereby output and assemble a plurality of phrases in a natural language that describe the structured data set based on the points that support the characterizations represented by the selected candidate angle data structures.

12. The apparatus of claim 11 wherein the processor is further configured to render the narrative story text for display based on a recursive traversal of the ordered element list data structure and an application of the recursively traversed elements from the ordered element list data structure to the invoked phrasal generation routine to thereby output and assemble a plurality of phrases in a natural language that describe the structured data set based on the points that support the characterizations represented by the selected candidate angle data structures.

13. The apparatus of claim 11 wherein the processor is further configured to (1) filter the selected candidate angle data structures based at least in part on the importance values associated with the selected candidate angle data structures to yield a set of filtered angle data structures, and (2) perform the order determination with respect to the filtered angle data structures.

14. The apparatus of claim 1 wherein the structured data set

Assignees

• Narrative Science Inc

Inventors

• Birnbaum Lawrence A
• Hammond Kristian J
• Allen Nicholas D
• Templon John R

Classifications

• G06F40/166

  Editing, e.g. inserting or deleting · CPC title

• G06F40/289

  Phrasal analysis, e.g. finite state techniques or chunking · CPC title

• G06F16/22

  Indexing; Data structures therefor; Storage structures · CPC title

• G06F16/2365

  Ensuring data consistency and integrity · CPC title

• G06Q10/10

  Office automation; Time management · CPC title