Cyberpersonalities in artificial reality

• US 8,719,200 B2
• Filed: 06/29/2007
• Issued: 05/06/2014
• Est. Priority Date: 06/29/2006
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: Mycybertwin Group Pty Limited
• Inventors: Zakos, John, Beilby, Liesl Jane
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): Seck, Ababacar
• Application Number:
  US12/306,563
• Publication Number:
  US 20090254417A1
• Time in Patent Office: 2,503 Days
• Field of Search: 706/45
• US Class Current: 706/45
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/004 Artificial life, i.e. computing arrangements simulating life
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06Q 30/0277 Online advertisement
• CPC Class Codes: G09B 5/00 Electrically-operated educational appliances working with questions and answers G09B7/00; simulators G09B9/00; advertising or displaying in general G09F
• CPC Class Codes: G09B 7/00 Electrically-operated teaching apparatus or devices working with questions and answers mechanically operated G09B3/00; computing arrangements G06F

Abstract

The invention concerns cyberpersonalities, including their and varied use in artificial reality. A cyberpersonality is comprised of a base personality (12). The base personality (12) is selected from a set of base personalities, each one representing the personality of a theoretical person. The cyberpersonality also includes a dynamic personality (14) that reflects the actual person (real or company) that the cyberpersonality is meant to represent and is able to learn. Information contained in the base (12) and dynamic (14) personality can be used to allow the person that the cyberpersonality mimics to interact in the artificial reality without direct control. The cybersonality can chat with third parties, including asking questions and answering questions, so as to learn more about each other. Other uses are related to searching, advertising and direct marketing.

Claims

1. A method of operating a computer system to provide a conversational agent to mimic a user, the method comprising:

providing a database accessible by the computer system, the database including a base personality component selected from a plurality of personalities, and a learning personality component capable of learning from content via at least only one modality attributable to a user so as to also mimic the user;

receiving in the computer system an input message from at least one of a client or an audience;

processing in the computer system content of the input message to determine one or more candidate responses corresponding with at least one of the base personality component or the learning personality component, wherein the learning personality component capable to perform additional learning from additional content via the at least only one modality attributable to the user to determine one or more candidate responses; and

generating in the computer system an output response message directed to said at least one of a client or an audience, and comprising a selected one of the candidate responses.

2. The method of claim 1 comprising selecting the base personality component from a database which includes one or more personalities, based upon one or more personality models.

3. The method of claim 1 comprising selecting the base personality component from a database which includes a plurality of personality characteristics, wherein the base personality component combines two or more personality models.

4. The method of claim 2 comprising identifying the one or more personality models of the base personality.

5. The method of claim 4 wherein the identifying the one or more personality models employs one or more of:

user responses to direct questions;

user expression derived from online activity of the user;

or group associations of the user.

6. The method of claim 2 wherein the database further includes one or more base personality typologies, and the base personality component comprises a combination of one or more of the personality models into one of the base personality typologies.

7. The method of claim 1, further comprising the learning personality component storing knowledge for subsequent response generation, wherein the knowledge is generated from processing additional content via at least only one modality attributable to a model user.

8. The method of claim 7 wherein the content comprises one or more of inputs and/or responses of the model user to the conversational agent;

questions and/or corresponding answers provided by the model user;

remarks and/or statements provided by the model user;

inputs and/or responses of the model user recorded from an exchange with another conversational agent; and

/or feedback provided via one or more interactions with the conversational agent by other users familiar with the model user.

9. The method of claim 7 wherein the processing comprises using one or more knowledge extraction processes.

10. The method of claim 1 further comprising:

monitoring sources of user content;

retrieving user content identified in the monitored sources;

extracting knowledge pertaining to the user from the retrieved content; and

using the extracted knowledge for subsequent response generation.

11. The method of claim 10 wherein the sources of user content comprise one or more of:

instant messaging content;

email content;

speech-to-text content;

blog content; and

/or online content identified as of-interest to the user.

12. The method of claim 1 further comprising the learning personality component storing knowledge for subsequent response generation wherein the knowledge is generated from processing additional content via at least only one modality attributable to the user;

wherein the learning personality component stores knowledge for subsequent response generation by;

processing additional content via the at least only one modality attributable to the user to generate usable knowledge;

storing the usable knowledge in a knowledge base; and

using the knowledge base to process content of the input message to determine one or more candidate responses.

13. The method of claim 12, wherein the processing further comprises:

parsing the additional content to identify at least one response;

extracting from the response terms and relationships between terms; and

storing extracted term relationships in a term matrix within the knowledge base.

14. The method of claim 13 further comprising storing unique responses in a response database within the knowledge base for subsequent retrieval as candidate responses to input messages.

15. The method of claim 12 wherein the processing further comprises:

parsing the additional content to identify one or more input-response pairs;

extracting one or more input-response pairs to generate one or more knowledge packets from the additional content; and

storing the knowledge packets in a knowledge matrix within the knowledge base.

16. The method of claim 15 wherein a knowledge packets comprises pairings of substrings comprising an input and a response.

17. The method of claim 16 wherein in a pairing of substrings in a knowledge packet, a substring comprises a three-word triplets.

18. The method of claim 1 wherein the processing content of the input message further comprises:

determining whether any candidate responses are available corresponding with the learning personality component and/or whether any candidate responses are available corresponding with the base personality component;

ranking available candidate responses substantially according to a relationship with the input message; and

selecting a response from the candidate responses based substantially upon the ranking.

19. The method of claim 18

wherein the database includes a conversational thesaurus comprising corresponding entries with a plurality of concepts;

processing the input message to identify concepts appearing within the conversational thesaurus; and

, in the event that a concept is identified, determining a candidate response associated with the identified concept.

20. A computer-implemented system for providing a conversational agent to mimic a user, the system comprising:

a computer system; and

a database which comprises a base personality component selected from a plurality of personalities to mimic the user, and a learning personality component capable of learning from content via at least only one modality attributable to the user so as to also mimic the user;

an interface to receive an input message from at least one of a client or an audience user;

a chat engine to process content of the input message to determine one or more candidate responses corresponding with at least one of the base personality component or the learning personality component wherein the learning personality component capable to perform additional learning from additional content via the at least only one modality attributable to the user to determine one or more candidate responses; and

an output interface to output a response message directed to the at least one of a client or an audience, and comprising a selected one of the candidate responses.

21. The system of claim 20 comprising:

a knowledge base; and

a knowledge learning engine to process content via the at least only one modality attributable to the user to store knowledge in the knowledge base generated from processing the content, wherein the chat engine to access the knowledge base to determine one or more candidate responses.

22. The system of claim 21 comprising:

a parser to parse additional content via the at least only one modality attributable to the user to identify at least one response; and

a response learner to extract terms and relationships between terms, and to store the extracted term relationships in a term matrix within the knowledge base.

23. The system of claim 22 wherein the response learner is further to store unique responses in a response database within the knowledge base for subsequent retrieval as candidate responses to input messages.

24. The system of claim 21 comprising:

a parser to parse additional content via the at least only one modality attributable to the user to identify one or more inputs-response pairs;

an input response learner to extract one or more knowledge packets from the input and from the response, and to store the knowledge packets in a knowledge matrix within the knowledge base.

25. The system of claim 24 wherein a knowledge packet comprises pairings of substrings comprising an input and a response.

26. The system of claim 25 wherein in a pairing of substrings in a knowledge packet, a substring comprises a three-word triplets.

27. The system of claim 20 wherein the chat engine is further to:

determine whether any candidate responses are available corresponding with the learning personality component and/or whether any candidate responses are available corresponding with the base personality component;

rank available candidate responses substantially according to a relationship with the input message; and

select a response from the candidate responses substantially based on the ranking.

28. The system of claim 20 further comprising a monitor to identify and retrieve additional content via the at least only one modality attributable to the user for use by the learning personality component.

29. The system of claim 28 wherein the monitor comprises a software component residing on a client computer located remotely from the chat engine, which also comprises a software component residing on a server, the monitor to upload retrieved content to the server via a data network connection.

30. The system of claim 28 wherein additional content comprises one or more of:

instant messaging content;

email content;

speech-to-text content;

blog content; and

/or online content identified as of interest to the user.

32. The apparatus of claim 20, wherein the learning personality component capable to perform additional learning from additional content via the at least only one modality attributable to the user to determine one or more candidate responses is capable to perform additional learning via learning which terms are related and extracting that knowledge from the additional content.

31. A non-transitory and tangible computer-readable storage medium having program instructions stored thereon which, when executed by a processor, cause the processor to implement a method comprising:

accessing a database associated with the computer system, the database containing a cyberpersonality data structure which comprises a base personality component selected from a plurality of predefined personalities, and a learning personality component configured to acquire and store knowledge for subsequent response generation based upon user inputs and user related information, wherein the base personality component and learning personality component are each configured to provide responses to input messages from at least one of a client or an audience;

receiving an input message from at least one of a client or an audience;

processing content of said input message with the cyberpersonality data structure to determine one or more candidate responses corresponding with at least one of the base personality component or the learning personality component of the cyberpersonality data structure; and

generating an output response message directed to at least one of a client or an audience, and comprising a selected one of said candidate responses.