EMPATHY INJECTION FOR QUESTION-ANSWERING SYSTEMS

• US 20150100521A1
• Filed: 10/09/2013
• Published: 04/09/2015
• Est. Priority Date: 10/09/2013
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Kozloski, James R., Murdock, James W. IV, Pickover, Clifford A., Walker, George F.
• Granted Patent: US 9,298,766 B2
• Time in Patent Office: Days
• US Class Current: 706/11
• CPC Class Codes: G06F 16/24 Querying
• CPC Class Codes: G06F 16/24522 Translation of natural language queries to structured queries
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/248 Presentation of query results
• CPC Class Codes: G06F 16/3331 Query processing
• CPC Class Codes: G06F 16/337 Profile generation, learning or modification
• CPC Class Codes: G06F 3/0484 for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range
• CPC Class Codes: G06F 40/205 Parsing
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G16H 10/20 for electronic clinical trials or questionnaires
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/50 for simulation or modelling of medical disorders

Abstract

Exemplary methods and devices herein receive an inquiry and automatically analyze words used in the inquiry, potential answers, and data maintained by evidence sources using the computerized device to determine the sensitivity level associated with the inquiry. The sensitivity level associated with the inquiry represents an emotional and cognitive state of the user. Such methods and devices automatically generate at least one follow-up question based on the sensitivity level associated with the inquiry and receive a follow-up response into the computerized device in response to the follow-up question(s). The methods and devices also automatically produce scores for the potential answers using the computerized device based on the inquiry, the follow-up responses, and ratings of the evidence sources. Following this, these methods and devices automatically generate output answers to the inquiry based on the sensitivity level associated with the inquiry using the computerized device.

Claims

1. A method comprising:

receiving an inquiry from a user into a computerized device;

automatically performing an initial analysis of said inquiry using said computerized device to produce potential answers to said inquiry from evidence sources;

automatically analyzing words used in said inquiry, said potential answers, and data maintained by said evidence sources using said computerized device to determine a sensitivity level associated with said inquiry, said sensitivity level associated with said inquiry representing an emotional and cognitive state of said user as automatically determined by said computerized device; and

automatically generating at least one response to said inquiry, said response being one of relatively more empathetic and relatively less empathetic based on said sensitivity level associated with said inquiry.

2. The method according to claim 1, said response comprising at least one follow-up question, said method further comprising:

automatically outputting said at least one follow-up question using said computerized device in response to said inquiry;

receiving a follow-up response into said computerized device in response to said at least one follow-up question;

automatically producing scores for said potential answers using said computerized device based on said inquiry, said follow-up response, and ratings of said evidence sources;

automatically generating output answers to said inquiry based on said sensitivity level associated with said inquiry using said computerized device to refine said potential answers, said generating of said output answers generating one of relatively more empathetic output answers and relatively less empathetic output answers based on said sensitivity level associated with said inquiry; and

automatically outputting said output answers, said scores, and links to said evidence sources from said computerized device.

3. The method according to claim 2, said relatively more empathetic output answers and said relatively less empathetic output answers being distinguished from each other by an amount of information presented and a word choice when presenting information.

4. The method according to claim 1, said method determining said sensitivity level associated with said inquiry based on at least one of:

a manually generated list of responses that are known to be associated with various emotional states in response to various types of questions;

an automatically mined list of terms that occur disproportionately in said response indicating negative sentiment;

a structured knowledge resource containing relationships between entities and emotionally significant phenomena;

a manually generated list of keywords that tend to co-occur in said response that suggest a negative emotional state and keeping a count of how many times said keywords appear in said response;

a precompiled model of the emotional states suggested by said response; and

a plurality of models of emotional states suggested by said response based on a one or more patient-specific features.

5. The method according to claim 1, further comprising:

outputting said emotional and cognitive state of said user as automatically determined by said computerized device to individuals other than said user; and

making recommendations to said individuals based on said emotional and cognitive state of said user.

6. The method according to claim 1, said analyzing words used in said inquiry, said potential answers, and data maintained by said evidence sources comprising at least one of:

automatically identifying words associated with a specific emotional and cognitive state;

automatically identifying sentence structure associated with a specific emotional and cognitive state;

automatically identifying relationships between data maintained by said evidence sources and emotionally significant phenomena; and

automatically detecting negative sentiment associated with one of said potential answers.

7. A method comprising:

receiving an inquiry from a user into a computerized device;

automatically performing an initial analysis of said inquiry using said computerized device to produce potential answers to said inquiry from evidence sources;

automatically analyzing words used in said inquiry, said potential answers, and data maintained by said evidence sources using said computerized device to determine a sensitivity level associated with said inquiry, said sensitivity level associated with said inquiry representing an emotional and cognitive state of said user as automatically determined by said computerized device;

automatically generating at least one follow-up question based on said sensitivity level associated with said inquiry using said computerized device to refine said potential answers, said generating of said follow-up question generating one of relatively more empathetic follow-up questions and relatively less empathetic follow-up questions based on said sensitivity level associated with said inquiry;

automatically outputting said at least one follow-up question using said computerized device in response to said inquiry;

receiving a follow-up response into said computerized device in response to said at least one follow-up question;

automatically analyzing words used in said follow-up response using said computerized device to refine said sensitivity level associated with said inquiry;

automatically producing scores for said potential answers using said computerized device based on said inquiry, said follow-up response, and ratings of said evidence sources;

automatically generating output answers to said inquiry based on said sensitivity level associated with said inquiry using said computerized device to refine said potential answers, said generating of said output answers generating one of relatively more empathetic output answers and relatively less empathetic output answers based on said sensitivity level associated with said inquiry; and

automatically outputting said output answers, said scores, and links to said evidence sources from said computerized device.

8. The method according to claim 7, further comprising automatically obtaining biometric information from biometric devices operatively connected to said computerized device, said computerized device using said biometric information when determining said sensitivity level associated with said inquiry.

9. The method according to claim 7, said relatively more empathetic follow-up questions and said relatively less empathetic follow-up questions being distinguished from each other by an amount of information presented and a word choice when presenting information;

and

said relatively more empathetic output answers and said relatively less empathetic output answers being distinguished from each other by an amount of information presented and a word choice when presenting information.

10. The method according to claim 7, said outputting of said output answers comprising automatically switching to a virtual universe and using an avatar, said avatar changing appearance and mannerisms based on said sensitivity level associated with said inquiry.

11. The method according to claim 7, further comprising:

outputting said emotional and cognitive state of said user as automatically determined by said computerized device to individuals other than said user;

using an active learning component to access specific information from said individuals; and

making recommendations to said individuals based on said emotional and cognitive state of said user and output from said active learning component.

12. The method according to claim 7, said analyzing words used in said inquiry, said potential answers, and data maintained by said evidence sources comprising at least one of:

automatically identifying words associated with a specific emotional and cognitive state;

automatically identifying sentence structure associated with a specific emotional and cognitive state;

automatically identifying relationships between data maintained by said evidence sources and emotionally significant phenomena; and

automatically detecting negative sentiment associated with one of said potential answers.

13-18. -18. (canceled)

19. A method comprising:

receiving an inquiry from a user into a computerized device;

automatically performing an initial analysis of said inquiry using said computerized device to produce potential answers to said inquiry from evidence sources;

automatically analyzing words used in said inquiry, said potential answers, and data maintained by said evidence sources using said computerized device to determine a sensitivity level associated with said inquiry, said sensitivity level associated with said inquiry representing an emotional and cognitive state of said user as automatically determined by said computerized device;

automatically generating at least one follow-up question based on said sensitivity level associated with said inquiry using said computerized device to refine said potential answers;

automatically outputting said at least one follow-up question using said computerized device in response to said inquiry;

receiving a follow-up response into said computerized device in response to said at least one follow-up question;

automatically analyzing words used in said follow-up response using said computerized device to refine said sensitivity level associated with said inquiry;

automatically producing scores for said potential answers using said computerized device based on said inquiry, said follow-up response, and ratings of said evidence sources; and

automatically generating output answers to said inquiry based on said sensitivity level associated with said inquiry using said computerized device to refine said potential answers, said generating of said output answers generating one of relatively more empathetic output answers and relatively less empathetic output answers based on said sensitivity level associated with said inquiry.

20. The method according to claim 19, further comprising automatically obtaining biometric information from biometric devices operatively connected to said computerized device, said computerized device using said biometric information when determining said sensitivity level associated with said inquiry.

21. The method according to claim 19, said relatively more empathetic output answers and said relatively less empathetic output answers being distinguished from each other by an amount of information presented and a word choice when presenting information.

22. The method according to claim 19, said outputting of said output answers comprising automatically switching to a virtual universe and using an avatar, said avatar changing appearance and mannerisms based on said sensitivity level associated with said inquiry.

23. The method according to claim 19, further comprising:

outputting said emotional and cognitive state of said user as automatically determined by said computerized device to individuals other than said user;

using an active learning component to access specific information from said individuals; and

making recommendations to said individuals based on said emotional and cognitive state of said user and output from said active learning component.

24. The method according to claim 19, said analyzing words used in said inquiry, said potential answers, and data maintained by said evidence sources comprising at least one of:

automatically identifying words associated with a specific emotional and cognitive state;

automatically identifying sentence structure associated with a specific emotional and cognitive state;

automatically identifying relationships between data maintained by said evidence sources and emotionally significant phenomena; and

automatically detecting negative sentiment associated with one of said potential answers.