Dialogue strategies

• US 20070106628A1
• Filed: 09/29/2005
• Published: 05/10/2007
• Est. Priority Date: 09/29/2005
• Status: Abandoned Application

• Current Assignee: Conopco (Unilever PLC)
• Original Assignee: Conopco (Unilever PLC)
• Inventors: Adjali, Iqbal, Hurling, Robert, De Boni, Marco, Dias, Malcolm, Bataveljic, Ogi
• Application Number:
  US11/238,518
• Publication Number:
  US 20070106628A1
• Time in Patent Office: Days
• US Class Current: 706/45
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models

Abstract

A human-computer interface for automated adaptive persuasion dialogue and a method of operating such an interface. The method comprising presenting a user with a series of decision points, each requiring the user to select one of a plurality of possible decision options; presenting the user with at least one persuasion message corresponding to each of the possible decision options, each persuasion message being selected according to one of a plurality of different persuasion strategies and receiving the user selection of one of the possible decision options. The presenting and receiving steps are then repeated to obtain user selected decision options for decision points based on a plurality of persuasion strategies to determine a persuasion strategy that is optimised for the user, allowing subsequent persuasion messages to be delivered to the user based on the optimised persuasion strategy.

Claims

1. A method of operating a human-computer interface, comprising:

(a) presenting a user with a series of decision points, each decision point requiring the user to select one of a corresponding plurality of possible decision options;

(b) presenting the user with at least one persuasion message for each of the possible decision options, to persuade the user to select one of the decision options over each of the others at the time of presenting each decision point, each persuasion message being selected according to one of a plurality of different persuasion strategies;

(c) receiving the user selection of one of the possible decision options;

(d) repeating steps (a) to (c) to obtain user selected decision options for decision points based on a plurality of persuasion strategies to determine a persuasion strategy that is optimised for the user; and

(e) subsequently delivering persuasion messages to the user based on the optimised persuasion strategy.

2. The method of claim 1, in which step (d) is followed by a test or refinement phase, comprising the steps of:

(i) presenting the user with a second series of decision points, each decision point requiring the user to select one of a plurality of possible decision options;

(ii) presenting the user with at least one persuasion message corresponding to each one of the possible options, in which the persuasion message for one option is selected according to the optimised persuasion strategy and the persuasion message for another option is selected according to a non-optimised persuasion strategy;

(iii) receiving the user selection of one of the possible decision options;

(iv) repeating steps (i) to (iii) to determine a degree of accuracy of the optimised persuasion strategy; and

(v) adapting the optimised persuasion strategy

3. The method of claim 2 in which the steps (i) to (v) are performed during, or interspersed with, persuasion messages delivered in step (e).

4. The method of claim 1, further comprising generating the at least one persuasion message by:

choosing one of the plurality of different persuasion strategies;

adapting a content of a message template in accordance with the chosen persuasion strategy.

5. The method of claim 1, further comprising generating the at least one persuasion message by applying a natural language generation function to a message template to adapt a content of the template in accordance with one of the plurality of different persuasion strategies.

6. The method of claim 1, wherein determining an optimised persuasion strategy includes:

determining a strength of association between the user and each of the plurality of persuasion strategies; and

weighting the strengths of association based on the selected decision options, such that the association having the greatest weight indicates which persuasion strategy is optimum for the user.

7. The method of claim 6, wherein determining a strength of association includes assessing the probability of success of each of the plurality of persuasion strategies with the user.

8. The method of claim 6, wherein determining the optimised persuasion strategy further includes:

updating the weights of the strengths of association after each selected decision option is received from the user.

9. The method of claim 8, wherein the updating is based on a Hebbian reinforcement rule.

10. The method of claim 1, further comprising:

receiving real-time data relating to physical attributes of the user; and

using the data relating to the physical attributes in conjunction with the user selected decision options in determining a persuasion strategy that is optimised for the user.

11. The method of claim 1, further comprising:

detecting a location of the interface; and

modifying a content of a persuasion message as a function of the detected location.

12. The method of claim 1, wherein presenting and/or delivering the persuasion messages includes presenting the messages in one or more of the following formats:

textual, pictorial, graphical and audio.

13. The method of claim 1, wherein the plurality of different persuasion strategies is based on a Cialdini persuasion framework.

14. A human-computer interface for adaptive persuasion dialogue, comprising:

(a) means for presenting a user with a series of decision points, each decision point requiring the user to select one of a plurality of possible decision options;

(b) means for presenting the user with at least one persuasion message for each of the possible decision options to persuade the user to select one of the decision options over each of the others at the time of presenting each decision point, each persuasion message being selected according to one of a plurality of different persuasion strategies;

(c) means for receiving the user selection of one of the possible decision options;

(d) means for determining a persuasion strategy that is optimised for the user by repeating the presenting and receiving steps in (a) to (c) to obtain user selected decision options for decision points based on a plurality of persuasion strategies; and

(e) means for subsequently delivering persuasion messages to the user based on the optimised persuasion strategy.

15. The interface of claim 14, wherein the interface includes a test or refinement phase which follows the step of determining an optimised persuasion strategy, the interface further comprising:

(i) means for presenting the user with a second series of decision points, each decision point requiring the user to select one of a plurality of possible decision options;

(ii) means for presenting the user with at least one persuasion message corresponding to each one of the possible options, in which the persuasion message for one option is selected according to the optimised persuasion strategy and the persuasion message for another option is selected according to a non-optimised persuasion strategy;

(iii) means for receiving the user selection of one of the possible decision options;

(iv) means for determining a degree of accuracy of the optimised persuasion strategy by repeating the presenting and receiving steps of (i) to (iii); and

(v) means for adapting the optimised persuasion strategy.

16. The interface of claim 14, further comprising one or more biometric sensors for determining physical attributes of the user.

17. The interface of claim 14, further comprising means for detecting a location of the interface.