Causal modeling and attribution

• US 10,949,753 B2
• Filed: 04/03/2014
• Issued: 03/16/2021
• Est. Priority Date: 04/03/2014
• Status: Active Grant

• Current Assignee: Adobe Inc.
• Original Assignee: Adobe Inc.
• Inventors: Demiralp, Emre, Chang, Walter W.
• Primary Examiner(s): Gonzales, Vincent
• Application Number:
  US14/244,755
• Publication Number:
  US 20150286928A1
• Time in Patent Office: 2,539 Days
• Field of Search: None
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation

Abstract

In techniques for causal modeling and attribution, a causal modeling application implements a dynamical causal modeling framework. Input data is received as a representation of communications between users, such as social media interactions between social media users, and causal relationships between the users can be determined based in part on the input data that represents the communications. Influence variables, such as exogenous variables and/or endogenous variables, can also be determined that influence the causal relationships between the users. A causal relationships model is generated based on the influence variables and the causal relationships between the users, where the causal relationships model is representative of causality, influence, and attribution between the users.

Claims

1. A method, comprising:

receiving, by a computing device, communications between users that are modeled as text data, the text data including a sentiment expressed by one or more of the users about a subject in the communications;

analyzing, by the computing device, the text data to identify the sentiment about the subject, the analyzing includes calculating a weighted average of one or more sentiment scores associated with the sentiment of the subject in the communications;

generating, by the computing device, input data based on the weighted average of the one or more sentiment scores associated with the sentiment of the subject in the communications;

receiving, by a computing device, the input data as a representation of communications between users of social media;

determining, by the computing device, causal relationships between the users based in part on the input data and simultaneous modeling of one or more influence variables such that the simultaneous modeling incorporates random fluctuations associated with the one or more influence variables;

determining, by the computing device, one or more influence variables from the one or more influence variables that influence the causal relationships between the users, the one or more influence variables including one or more endogenous variables and one or more exogenous variables, the one or more exogenous variables moderating influence of the one or more endogenous variables on the causal relationships between the users;

generating, by the computing device, a causal relationships model based on the influence variables and the causal relationships between the users; and

controlling, by the computing device, an instance of content based on the causal relationships model.

2. The method as recited in claim 1, further comprising quantifying, by the computing device, the causal relationships with influence scores that indicate a degree to which one or more of the users influence a causal relationship with another of the users.

3. The method as recited in claim 1, wherein the causal relationships between the users are non-symmetric, and the causal relationships model is further generated to represent the non-symmetric causal relationships.

4. The method as recited in claim 1, wherein determining the causal relationships between the users further comprises:

applying, by the computing device, a dynamical causal modeling framework to the causal relationships.

5. The method as recited in claim 4, wherein the dynamical causal modeling framework accounts for hysteresis corrections associated with one or more of the users.

6. The method as recited in claim 1, wherein the generated input data comprises vector space representations of the communications that are represented as values.

7. The method as recited in claim 1, wherein the text data is received as part-of-speech information that includes one or more of noun expressions, verb expressions, and tagged parts-of-speech in the communications.

8. The method as recited in claim 1, wherein the causal relationships are representative of the non-symmetric influence and non-symmetric attribution between the users.

9. The method as recited in claim 1, wherein the one or more exogenous variables are independent variables that influence the causal relationships between the users without being affected by feedback from the users.

10. The method as recited in claim 1, wherein the one or more endogenous variables are dependent variables that change based on feedback from the users.

11. A computing device comprising:

a memory configured to maintain input data that is received as a representation of social media interactions between users of social media; and

a processor system to implement a causal modeling application that applies a dynamical causal modeling framework that is configured to;

receive communications between users that are modeled as text data, the text data including a sentiment expressed by one or more of the users about a subject in the communications;

analyze the text data to identify the sentiment about the subject, the analyzing includes calculating a weighted average of one more sentiment scores associated with the sentiment of the subject in the communications;

generate input data based on the weighted average of the one or more sentiment scores associated with the sentiment of the subject in the communications;

determine causal relationships between the users based in part on the input data that represents the social media interactions and simultaneous modeling of one or more influence variables such that the simultaneous modeling incorporates random fluctuations associated with the one or more influence variables;

determine one or more influence variables from the one or more influence variables that influence the causal relationships between the users, the one or more influence variables including one or more endogenous variables and one or more exogenous variables, the one or more exogenous variables moderating influence of the endogenous variables on the causal relationships between the users;

generate a causal relationships model based on the influence variables and the causal relationships between the users; and

controlling an instance of content based on the causal relationships model.

12. The computing device as recited in claim 11, wherein the dynamical causal modeling framework is configured to quantify the causal relationships with influence scores that each indicate a degree to which one or more of the users influence a causal relationship with another of the users.

13. The computing device as recited in claim 11, wherein the causal relationships between the users are non-symmetric, and the causal relationships model is representative of the non-symmetric causal relationships, influence, and attribution between the users.

14. The computing device as recited in claim 11, wherein the input data that represents the social media interactions between the users comprises vector space representations of the social media interactions that are represented as values.

15. A computer-readable storage media comprising a causal modeling application stored as instructions that are executable and, responsive to execution of the instructions by a computing device, the computing device performs operations comprising:

receiving communications between users that are modeled as text data, the text data including a sentiment expressed by one or more of the users about a subject in the communications;

analyzing the text data to identify the sentiment about the subject, the analyzing includes calculating a weighted average of one or more sentiment scores associated with the sentiment of the subject in the communications;

generating input data based on the weighted average of the one or more sentiment scores associated with the sentiment of the subject in the communications;

receiving input data as a representation of communications between users of social media;

determining causal relationships between the users based in part on the input data that represents the communications and simultaneously modeling of one or more influence variables such that the simultaneous modeling incorporates random fluctuations associated with the one or more influence variables;

determining one or more influence variables from the one or more influence variables that influence the causal relationships between the users, the influence variables including one or more exogenous variables and one or more endogenous variables, the one or more exogenous variables moderating influence of the one or more endogenous variables on the causal relationships between the users;

generating a causal relationships model based on the one or more influence variables and the causal relationships between the users; and

controlling an instance of content based on the causal relationships model.

16. The computer-readable storage media as recited in claim 15, wherein determining the causal relationships between the users further comprises applying a dynamic causal modeling framework to the causal relationships.

17. The computer-readable storage media as recited in claim 16, wherein the dynamic causal modeling framework accounts for hysteresis corrections associated with one or more of the users.

18. The computer-readable storage media as recited in claim 15 that comprises the causal modeling application stored as instructions that are executed, and responsive to execution of the instructions, the computing device is further configured to generate the input data as the representation of the communications between the users, the input data comprising vector space representations of the communications that are represented as values.

19. The computer-readable storage media as recited in claim 15, wherein the text data is received as part-of-speech information that includes one or more of noun expressions, verb expressions, and tagged parts-of-speech in the communications.

20. The computer-readable storage media as recited in claim 15, wherein the causal relationships are representative of the non-symmetric influence and non-symmetric attribution between the users.