System and method for contextual analysis

Patent metadata

Field	Value
Publication number	US-9524464-B2
Application number	US-201414198212-A
Country	US
Kind code	B2
Filing date	Mar 5, 2014
Priority date	Mar 5, 2013
Publication date	Dec 20, 2016
Grant date	Dec 20, 2016

Abstract

Computerized contextual analysis systems and methods suitable for monitoring situations, regions, and groups characterized by volatility and uncertainty are provided. Via use of exemplary systems and methods, decision makers, for example politicians, warfighters, and analysts can gain insight into the cultures, attitudes, events, and relationships that may impact their missions.

First claim

What is claimed is:

1. A method for contextual analysis, comprising:
obtaining, via a computer system configured with a software package for contextual analysis, representative communications associated with a sociopolitical concept;
assessing, via the software package for contextual analysis, the representative communications in order to form a contextual assessment; and
providing, to a user of the software package for contextual analysis, the contextual assessment;
wherein the assessing comprises perspective mining utilizing a logistic regression of the form:

min

x

⁢

∑

i
=
1

m

⁢

⁢

w
i

⁢

log
⁡

(

1
+

exp
⁡

(

-

y
i

⁡

(

x
T

⁢

a
i

+
c

)

)

)

+

λ
⁢


x


1

wherein in is the number of documents in the representative communications;
x T is s the transpose of x;
c is the intercept used in the logistic regression;
a i is the vector representation of the i th document the representative communications;
w i is the weight assigned to the i th document (w i =1/m by default);
A=[a 1 , a 2 , . . . , a m ] is a document keyword matrix;
y i is the polarity of each document based upon the scale polarity of the actor that the document belongs to;
the unknown x 1 , the j-th element of x, is the weight for each keyword;
λ>0 is a regularization parameter that controls the sparsity of the solution; and
|x| 1 =Σ|x i | is 1-norm of the x vector.

2. The method of claim 1 , wherein the assessing further comprises at least one of:
multi-scale comparative predictive modeling;
micro-level analysis of individuals or groups;
generating an alert for a breaking event; or
neutral-to-extreme perspective selection.

3. The method of claim 1 , wherein the contextual assessment comprises at least one of:
identification of shifts of individuals from unaffiliated to affiliated with a non-state actor (NSA);
growth or shrinkage drivers of the NSA;
identification of an influential follower of the NSA;
information regarding the presence of a faction within the NSA;
identification of wedge-driving rumors or perspectives that undermine or strengthen group loyalties in the NSA; or
identification of a previously unknown NSA.

4. The method of claim 1 , wherein the assessing comprises generating a Gutmann pattern arising from the representative communications.

5. The method of claim 1 , wherein the assessing utilizes at least one of path analysis or structural equation modeling.

6. The method of claim 1 , further comprising utilizing, by the user, the contextual assessment to determine a military response to a non-state actor (NSA).

7. The method of claim 1 , further comprising utilizing, by the user, the contextual assessment to determine a political response to a non-state actor (NSA).

8. The method of claim 1 , further comprising classifying, via the software package for contextual analysis, at least a subset of the representative communications as being associated with a specific non-state actor (NSA).

9. The method of claim 1 , wherein the assessing comprises:
generating a traversal sequence for the representative communications;
forming a graph model associated with messengers of the representative communications; and
determining, via evaluation of the traversal sequence, a likelihood of opinion shift for at least one messenger, wherein the determining utilizes a scaled opinion of the messenger regarding at least one of the representative communications.

Assignees

• Davulcu Hasan
• Woodward Mark
• Corman Steven
• Ye Jieping
• Univ Arizona State

Inventors

• Davulcu Hasan
• Woodward Mark
• Corman Steven
• Ye Jieping

Classifications

• G06N5/02 Primary

  Knowledge representation; Symbolic representation · CPC title