Method of analyzing a scenario represented as elements of a tensor space, and scored using tensor operators

Patent metadata

Field	Value
Publication number	US-9208440-B2
Application number	US-201313904456-A
Country	US
Kind code	B2
Filing date	May 29, 2013
Priority date	May 29, 2012
Publication date	Dec 8, 2015
Grant date	Dec 8, 2015

Abstract

A mathematical model to represent multiple role events of a threat scenario, and a methodology to score the events given the data or evidence contained in a knowledge base is provided. The multiple role event can include any action encompassed by an independent clause having multiple syntactical roles (subject, object, location, etc. . . . ) and containing multiple entities (person A, person B, object C, place D, etc. . . . ). Multiple role events are represented as elements of a tensor space, and are scored using tensor operators built from the semantic graph associated with the knowledge base.

First claim

The invention claimed is:

1. A computer implemented method for scoring a relative likelihood that a hypothetical event in a scenario is to occur, the method comprising:
collecting narrative text communications, each communication including a group of words, regarding evidence related to the scenario to build a knowledge base;
representing the evidence as a semantic graph of the knowledge base;
building a symmetric weighted adjacency matrix representing an undirected weighted t…

Assignees

• Battelle Memorial Institute

Inventors

• Morara Michele
• Davis Mark D
• Rust Steve W

Classifications

• G06Q50/265

  Physics · mapped topic

• G06N5/02 Primary

  Physics · mapped topic

• G06Q50/26

  Physics · mapped topic