Method and system for integrating multimodal interpretations

• US 20060143216A1
• Filed: 12/23/2004
• Published: 06/29/2006
• Est. Priority Date: 12/23/2004
• Status: Abandoned Application

• Current Assignee: Motorola, Inc. (Motorola Solutions, Inc.)
• Original Assignee: Motorola, Inc. (Motorola Solutions, Inc.)
• Inventors: Anastasakos, Tasos, Gupta, Anurag K.
• Application Number:
  US11/021,407
• Publication Number:
  US 20060143216A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 40/35 Discourse or dialogue representation

Abstract

A method and a system for integrating multimodal interpretations (MMIs) in user inputs to a data processing system (100) are provided. The method includes generating (603) a set of MMIs, based on the user inputs collected during a turn. The set of MMIs includes at least semantic relationship function, the semantic relationship function denoting a semantic relationship between values determined from the user inputs entered through one or more of the at least one input modality, the semantic relationship function comprising at least one semantic operator. The method further includes generating (604) sets of joint MMIs. Each set of joint MMIs includes MMIs of semantically compatible types. The method further includes resolving (606) the reference values of the references of each set of joint MMIs to generate corresponding sets of reference resolved MMIs. The method further includes generating (608) an integrated MMI for each set of reference-resolved MMIs. The generation of the integrated MMI is carried out by using semantic operator based fusion.

Claims

1. A method for integrating multimodal interpretations (MMIs) generated from user inputs to a data processing system, the user inputs being entered through at least one input modality, the method comprising:

generating a set of MMIs based on the user inputs for a user turn, at least one MMI comprising at least one semantic relationship function, the semantic relationship function denoting a semantic relationship between values determined from the user inputs entered through one or more of the at least one input modality, the semantic relationship function comprising at least one semantic operator;

generating one or more sets of joint MMIs by using the generated set of MMIs, each set of joint MMIs comprising one or more MMIs of semantically compatible types;

resolving reference values within each set of joint MMIs to generate corresponding sets of reference resolved MMIs; and

generating an integrated MMI for each set of reference resolved MMIs by using semantic operator based fusion.

2. The method according to claim 1, wherein generating the integrated MMI by semantic operator base fusion comprises:

determining the type of an integrated MMI that will be generated, based on the concept level relationships of the MMIs in the corresponding set of reference resolved MMIs;

determining final values of features in the integrated MMI based on the semantic operators present in values of corresponding features in the corresponding set of reference resolved MMIs.

3. The method according to claim 1, wherein generating the integrated MMI by semantic operator based fusion comprises generating an integrated MMI of an aggregate type for a set of reference resolved MMIs when all MMIs in the corresponding set of reference resolved MMIs are of a same concept type, and include no semantic relationship function, and wherein all features in the corresponding set of reference resolved MMIs have non-empty values.

4. The method in accordance with claim 1 further comprising generating a Multimodal structure (MMFS) for each MMI in the generated set of MMIs, the MMFS comprising at least one semantic operator.

5. The method in accordance with claim 1 wherein, generating the set of MMIs comprises setting a value of an attribute present in the MMI to a semantic relationship function.

6. The method in accordance with claim 1 wherein the MMIs of semantically compatible types are related by a concept level relationship that is one of a ‘

is a kind of’ and

‘

is a part of’

relationship.

7. The method in accordance with claim 1 wherein the semantic relationship function is selected from the group consisting of ‘

and’

, ‘

or’

, ‘

not’

, ‘

list’

, ‘

aggregate’

, ‘

ambiguous’ and

‘

replace’

relationship functions.

8. The method in accordance with claim 1 wherein, the semantic relationship function is a content level relationship.

9. The method in accordance with claim 8 wherein, an MMI corresponding to the content-level relationship comprises at least one of a complimentary interpretation, a redundant interpretation, a logical interpretation and an overlaying interpretation.

10. The method in accordance with claim 2 wherein, each TFS comprises at least one attribute-value pair, each attribute being one of an atomic attribute and a complex attribute, the atomic attribute being one of a group consisting of an empty value, one or more atomic values and a semantic relationship function, and the complex attribute being one of a group consisting of an empty value, one or more complex values, a reference variable and a semantic relationship function.

11. A method for integrating multimodal interpretations (MMIs) generated from user inputs to a data processing system, the user inputs being entered through at least one input modality, the method comprising:

segmenting the user inputs by collecting sets of multimodal interpretations (MMIs) corresponding to the user inputs for a user turn, at least one MMI comprising at least one semantic relationship function the semantic relationship function denoting a semantic relationship between values determined from the user inputs entered through one or more of the at least one input modality, the semantic relationship function comprising at least one semantic operator;

semantically classifying the collected MMIs to generate sets of joint MMIs, wherein each set of joint MMIs comprises MMIs of semantically compatible types;

generating one or more sets of reference resolved MMIs by resolving the reference variables in the sets of joint MMIs, wherein resolving the reference variables refers to replacing each reference variable with a resolved value; and

integrating each set of reference resolved MMIs to generate a corresponding set of integrated MMIs by using semantic operator based fusion.

12. A data processing system for integrating multimodal interpretations (MMIs) generated from user inputs to the data processing system, the user inputs being entered through at least one input modality, the data processing system generating MMIs based on each user input, the data processing system comprising:

a segmentation module, the segmentation module collecting a set of MMIs corresponding to the user inputs for a user turn, at least one MMI comprising at least one semantic relationship function the semantic relationship function denoting a semantic relationship between values determined from the user inputs entered through one or more of the at least one input modality, the semantic relationship function comprising at least one semantic operator;

a semantic classifier, the semantic classifier semantically classifying the set of MMIs to generate sets of joint MMIs, wherein each set of joint MMIs comprises MMIs of semantically compatible types;

a reference resolution module, the reference resolution module generating one or more sets of reference resolved MMIs by resolving the reference variables in the sets of joint MMIs, wherein resolving the reference variables refers to replacing each reference variable with a resolved value; and

an integrator module, the integrator module integrating each set of reference resolved MMIs to generate a corresponding set of integrated MMIs by using semantic operator based fusion.

13. An electronic equipment that integrates multimodal interpretations (MMIs) generated from user inputs to a data processing system, the user inputs being entered through at least one input modality, the equipment comprising:

means for generating a set of MMIs based on the user inputs for a user turn, at least one MMI comprising at least one semantic relationship function, the semantic relationship function denoting a semantic relationship between values determined from the user inputs entered through one or more of the at least one input modality, the semantic relationship function comprising at least one semantic operator;

means for generating one or more sets of joint MMIs by using the generated set of MMIs, each set of joint MMIs comprising one or more MMIs of semantically compatible types;

means for resolving reference values within each set of joint MMIs to generate corresponding sets of reference resolved MMIs; and

means for generating an integrated MMI for each set of reference resolved MMIs by using semantic operator based fusion.

14. A computer program product for use with a computer, the computer program product comprising a computer usable medium having a computer readable program code embodied therein for integrating multimodal interpretations (MMIs) generated from user inputs to a data processing system, the user inputs being entered through at least one input modality, the computer program code including functions for:

generating a set of MMIs based on the user inputs for a user turn, at least one MMI comprising at least one semantic relationship function, the semantic relationship function denoting a semantic relationship between values determined from the user inputs entered through one or more of the at least one input modality, the semantic relationship function comprising at least one semantic operator;

generating one or more sets of joint MMIs by using the generated set of MMIs, each set of joint MMIs comprising one or more MMIs of semantically compatible types;

resolving reference values within each set of joint MMIs to generate corresponding sets of reference resolved MMIs; and

generating an integrated MMI for each set of reference resolved MMIs by using semantic operator based fusion.

15. A method for integrating multimodal interpretations (MMIs) generated from user inputs to a data processing system, the user inputs being entered through at least one input modality, the method comprising:

generating a set of MMIs based on the user inputs for a user turn;

generating one or more sets of joint MMIs by using the generated set of MMIs, each set of joint MMIs comprising one or more MMIs of semantically compatible types;

resolving reference values within each set of joint MMIs to generate corresponding sets of reference resolved MMIs; and

generating an integrated MMI of an aggregate type for each set of reference resolved MMIs when all MMIs in a corresponding set of reference resolved MMIs are of a same concept type and wherein all features in the set of reference resolved MMIs have non-empty values.