Dialog system

• US 20020065651A1
• Filed: 09/18/2001
• Published: 05/30/2002
• Est. Priority Date: 09/20/2000
• Status: Abandoned Application

• Current Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Original Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Inventors: Souvignier, Bernd, Philips, Petra, Kellner, Andreas, Portele, Thomas
• Application Number:
  US09/954,657
• Publication Number:
  US 20020065651A1
• Time in Patent Office: Days
• US Class Current: 704/231
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/227 of the speaker; Human-factor methodology
• CPC Class Codes: G10L 2015/228 of application context

Abstract

The invention relates to a dialog system (1) which has a most comfortable and effective dialog structure for a user, comprising processing units for

automatic speech recognition (3),

natural language understanding (4),

defining system outputs in dependence on information (7) derived from user inputs,

generating acoustic and/or visual system outputs (9, 10, 11, 12),

deriving user models, while the user models contain details about the style of speech of user inputs and/or details about interactions in dialogs between users and the dialog system (1) and adaptation of contents and/or form of system outputs is provided in dependence on the user models.

Claims

1. A dialog system (1) comprising processing units for

automatic speech recognition (3),

natural language understanding (4),

defining system outputs in dependence on information (7) derived from user inputs,

generating acoustic and/or visual system outputs (9, 10, 11, 12),

deriving user models (22, 25), while the user models (22, 25) contain details about the style of speech of user inputs and/or details about interactions in dialogs between users and the dialog system (1) and adaptation of contents and/or form of system outputs is provided in dependence on the user models (22, 25).

2. A dialog system as claimed in claim 1, characterized

in that in addition to the input modality to use user inputs by means of speech, at least a further input modality is provided and

in that the user models (22, 25) contain details about the respective use of the various input modalities by the user.

3. A dialog system as claimed in claim 1 or 2, characterized

in that the user models (22, 25) contain estimates for the reliability of recognition results derived from user inputs.

4. A dialog system as claimed in claim 3, characterized

in that in dependence on the estimates, system responses are generated which prompt the respective user to use such input modalities for which high estimate values were determined and/or which prevent the respective user from using input modalities for which low reliability values were determined.

5. A dialog system as claimed in one of the claims 1 to 4, characterized

in that fixed models of user stereotypes (22) are used for forming the user models.

6. A dialog system as claimed in one of the claims 1 to 5, characterized

in that user models (25) are used which are continuously updated based on inputs of the respective user.

7. A method of operating a dialog system, in which processing units are used for

automatic speech recognition (3),

natural language understanding (4),

defining system outputs in dependence on information (7) derived from user inputs,

generating acoustic and/or visual system outputs (9, 10, 11, 12), and

deriving user models (13),

while the user models contain details about the style of speech of user inputs and/or indications about interactions in dialogs between users and the dialog system (1) and an adaptation of contents and/or form of system outputs is provided in dependence on the user models (22, 25).