Method and system for automating scenario planning

Patent metadata

Field	Value
Publication number	US-12456072-B2
Application number	US-202117207805-A
Country	US
Kind code	B2
Filing date	Mar 22, 2021
Priority date	Mar 22, 2021
Publication date	Oct 28, 2025
Grant date	Oct 28, 2025

Abstract

Embodiments are provided that relate to a computer system, a computer program product, and a computer-implemented method for automating scenario planning. Embodiments involve machine learning (ML) and an artificial intelligence (AI) planner to capture a general scenario planning (GSP) problem and provide a solution to the GSP problem in the form of trajectories.

First claim

What is claimed is:

1. A computer system comprising:
a processor operatively coupled to memory;
an artificial intelligence (AI) platform, operatively coupled to the processor, comprising:
a natural language (NL) module configured to interpret a received NL issue requiring decision-making, and to identify a corpus of text related to the interpreted NL issue;
a force module configured to process the identified corpus of text to derive a document set of forces, wherein the set of forces are derived automatically by the force module, wherein deriving the document set of forces includes the force module to automatically derive a set of initial forces, a set of implication forces, a set of selected forces, a set of indicator forces from the identified corpus of text, and a final set of forces based on a ratio of the set of initial forces and the set of final forces being less than a threshold;
a forces causal model (FCM) module configured to leverage the document set of forces and automatically construct a corresponding FCM, wherein the automatically constructed corresponding FCM is configured to mitigate bias in a solution, and the FCM maps a force to a vector that describe properties of a sequence from a first node to a second node in a graph diagram;
a machine learning (ML) module configured to leverage the corresponding FCM to:
construct a general scenario planning (GSP) problem;
translate the GSP problem into a planning problem to obtain a set of plans, wherein each plan of the set of plans includes a valid trajectory through the FCM, wherein the valid trajectory starts at an initial force of the set of initial forces and passes through at least one implication of the set of implications; and
provide the solution to the planning problem, the solution including computing one or more trajectories; and

a visual interface configured to present the solution as the one or more trajectories.

2. The computer system of claim 1 , wherein the one or more trajectories is the graph diagram containing one or more nodes and one or more edges, wherein each node represents a condition or event and each edge indicates a causal relationship between two nodes.

3. The computer system of claim 2 , further comprising the ML module to provide an explanation for the one or more nodes and the one or more edges, including link each node and edge to one or more documents of the identified corpus of text.

4. The computer system of claim 1 , wherein the solution includes a plurality of trajectories, and further comprising a cluster module operatively coupled to the ML module configured to cluster the plurality of trajectories using a clustering algorithm.

5. The computer system of claim 1 , wherein the received NL issue is one or more phrases defining one or more events or conditions requiring decision-making, and further comprising the NL module to automatically extract the one or more phrases from an input source.

6. The computer system of claim 1 , further comprising the ML module to:
define the GSP problem based on the FCM and the derived set of initial forces, set of implication forces, set of selected forces, and set of indicator forces;
leverage an artificial intelligence (AI) planner to generate a planning task, wherein output from the AI planner is the set of plans; and
compute the one or more trajectories for each plan in the set of plans, wherein each of the one or more trajectories is a sequence of forces.

7. A computer program product to support decision-making, the computer program product comprising
a computer readable storage medium or media; and
program code stored on the computer readable storage medium or media and executable by a computer processor that causes the processor to:
interpret a received natural language (NL) issue requiring decision-making, and to identify a corpus of text related to the interpreted NL issue;
process the identified corpus of text to derive a document set of forces, wherein the set of forces are derived automatically by a force module, wherein deriving the document set of forces includes the force module to automatically derive a set of initial forces, a set of implication forces, a set of selected forces, a set of indicator forces from the identified corpus of text, and a final set of forces based on a ratio of the set of initial forces and the set of final forces being less than a threshold;
leverage the document set of forces and automatically construct a corresponding forces causal model (FCM), wherein the automatically constructed corresponding FCM mitigate bias in a solution, and the FCM maps a force to a vector that describe properties of a sequence from a first node to a second node in a graph diagram;
employ machine learning (ML) to leverage the FCM, including:
construct a general scenario planning (GSP) problem;
translate the GSP problem into a planning problem to obtain a set of plans, wherein each plan of the set of plans includes a trajectory through the FCM wherein the trajectory starts at an initial force of the set of initial forces and passes through at least one implication of the set of implications; and
provide the solution to the planning problem, the solution including computing one or more trajectories; and

present the solution as the one or more trajectories via a visual interface.

8. The computer program product of claim 7 , wherein the one or more trajectories is the graph diagram containing one or more nodes and one or more edges, wherein each node represents a condition or event and each edge indicates a causal relationship between two nodes.

9. The computer program product of claim 8 , further comprising program code to provide an explanation for the one or more nodes and the one or more edges, including link each node and edge to one or more documents of the identified corpus of text.

10. The computer program product of claim 7 , wherein the solution includes a plurality of trajectories, and further comprising program code to cluster the plurality of trajectories using a clustering algorithm.

11. The computer program product of claim 7 , wherein the received NL issue is one or more phrases defining one or more events or conditions requiring decision-making, and further comprising program code to automatically extract the one or more phrases from an input source.

12. The computer program product of claim 7 , further comprising program code to:
define the GSP problem based on the FCM and the derived set of initial forces, set of implication forces, set of selected forces, and set of indicator forces;
leverage an artificial intelligence (AI) planner to generate a planning task, wherein output from the AI planner is the set of plans; and
compute the one or more trajectories for each plan in the set of plans, wherein each of the one or more trajectories is a sequence of forces.

13. A method of using a computing device to support decision-making, the method comprising:
interpreting a received natural language (NL) issue requiring decision-making, and identifying a corpus of text related to the interpreted NL issue;
processing the identified corpus of text and deriving a document set of forces, wherein the set of forces are derived automatically by a force module, wherein deriving the document set of forces includes the force module to automatically derive a set of initial forces, a set of implication forces, a set of selected forces, a set of indicator forces from the identified corpus of text, and a final set of forces based on a ratio of the set of initial forces and the set of final forces being less than a threshold;
leveraging the document set of forces and automatically constructi

Assignees

• IBM

Inventors

• Udrea Octavian
• Sohrabi Araghi Shirin
• Katz Michael
• Feblowitz Mark David
• Srinivas Kavitha
• Hassanzadeh Oktie

Classifications

• G06F40/20 Primary

  Natural language analysis (semantic analysis of natural language G06F40/30) · CPC title

• G06F40/289

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

• G06N20/00 Primary

  Machine learning · CPC title