Personalized user specific files for object recognition

• US 20070156682A1
• Filed: 12/28/2005
• Published: 07/05/2007
• Est. Priority Date: 12/28/2005
• Status: Abandoned Application

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Howell, David, Didcock, Clifford
• Application Number:
  US11/320,183
• Publication Number:
  US 20070156682A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules
• CPC Class Codes: G10L 2015/227 of the speaker; Human-factor methodology
• CPC Class Codes: G10L 2015/228 of application context
• CPC Class Codes: H04M 3/4936 Speech interaction details speech recognition per se G10L15/00

Abstract

A system for identifying an object corresponding to user input including a first grammar, a user specific file and a controller. The first grammar is a list of objects. The user specific file designates a subset of the list of objects in the first grammar which are related to the user. The controller receives the user input and compares the user input to the objects of at least the user specific file. The controller identifies to the user the object corresponding to the user input, giving preference to those objects designated in the user specific file.

Claims

1. A system for identifying an object corresponding to user input from a user, said system comprising:

a first grammar having a list of objects;

a user specific file designating a subset of the list of objects of the first grammar wherein the subset is related to the user;

a controller receiving user input and comparing the user input to the subset of objects of the first grammar and giving preference to the objects of the user specific file which correspond to the user input over objects of the first grammar which correspond to the user input wherein the controller identifies the object corresponding to the user input.

2. The system of claim 1 wherein the controller compares the user input to the objects in the first grammar to determine a probability that the user input corresponds to each object, increases or decreases the probability that the user input corresponds to each object according to any weighting for any object in the user specific file, and identifies the object with the highest probability as the object corresponding to the user input.

3. The system of claim 1 wherein the controller compares the user input to the objects in the first grammar to determine a probability that the user input corresponds to each object, compares the user input to objects in the user specific file to determine a probability that the user input corresponds to each object, and identifies the object with the highest probability as the object corresponding to the user input.

4. The system of claim 1 further comprising a plurality of grammars wherein each grammar is a subset of another and preference is given to the objects in each smaller grammar.

5. The system of claim 1 wherein the user specific file is a plurality of grammars wherein each of the plurality of grammars represents a varying degree of preference for the objects therein.

6. The system of claim 1 wherein the controller compares the user input to objects in the user specific file to determine a probability that the user input corresponds to each object, compares the user input to objects in the first grammar if a probability from comparing the user input to objects of the user specific file is not above a threshold, and identifies the object with the highest probability as the object corresponding to the user input.

7. The system of claim 1 wherein each of said objects comprises the name of a person and their associated nicknames.

8. The system of claim 1 wherein the system advertises at least one particular object by designating the particular object in the user specific file.

9. The system of claim 1 wherein the user specific file designates a first object as preferred for identification and wherein when the user input corresponds to both the first object and a second object, the controller identifies the first object to the user.

10. A user specific file for use by a user wherein the user specific file is based on a grammar resource including at least one of inboxes, electronic mail folders, directories, contacts list, calendars, organizational charts, advertising lists, user profile, and call logs, said user specific file comprising:

a list of objects selected from the grammar resource wherein the objects are selected based on a relationship of the object to the user; and

a weight associated with each object in the list as a function of the relationship of the object to the user.

11. The user specific file of claim 10 wherein the relationship of the object to the user is at least partially determined by at least one of:

how recently the object appears in the grammar resource;

the number of occurrences of the object in the grammar resource; and

what type of grammar resource the object occurs in.

12. The user specific file of claim 10 wherein each of said objects comprises the name of a person and their associated nicknames.

13. The user specific file of claim 10 wherein the list of objects selected from the grammar resource includes at least one object from an advertising list and the weight associated with the object from an advertising list is determined at least partially based on the advertising list.

14. A method of designating a file specific to a user based on a grammar resource including at least one of inboxes, electronic mail folders, directories, calendars, contacts list, organizational charts, advertising lists, user profile, and call logs of the user, wherein objects designated in said user specific file are given preference over other objects when determining an object corresponding to input from the user, said method comprising:

gathering statistics of usage by the user of objects in the grammar resource;

designating the objects having higher usage by the user based on the gathered usage information wherein the designated objects are given preference over undesignated objects when determining an object corresponding to input from the user.

15. The method of claim 14 wherein said gathering comprises:

searching the grammar resource for objects;

counting the occurrences of each object in each type of grammar resource; and

determining how recently the object occurred in the grammar resource.

16. The method of claim 14 wherein each of said objects comprises the name of a person and their associated nicknames.

17. The method of claim 14 wherein designating the objects having higher usage by the user comprises assigning a value to each object wherein the value represents a stronger preference for an object having higher usage information and a weaker preference for an object having lower usage information when determining an object corresponding to input from the user.

18. The method of claim 14 wherein designating the objects having higher usage by the user comprises sorting objects into bins according to similar usage information wherein bins having objects with higher usage information are preferred more than bins with lower usage information when determining an object corresponding to input from the user.

19. The method of claim 14 further comprising designating objects based on an advertising list wherein the designated objects are given preference over undesignated objects when determining an object corresponding to input from the user.

20. The method of claim 14 further comprising designating objects based on a user profile wherein the designated objects are given preference over undesignated objects when determining an object corresponding to input from the user.