Method for personalization of a service

• US 8,005,680 B2
• Filed: 11/21/2006
• Issued: 08/23/2011
• Est. Priority Date: 11/25/2005
• Status: Active Grant

• Current Assignee: Huawei Technologies Co., Ltd. (Huawei Investment & Holding Co., Ltd.)
• Original Assignee: Swisscom AG (Government of Switzerland)
• Inventors: Kommer, Robert Van
• Primary Examiner(s): ABEBE, DANIEL DEMELASH
• Application Number:
  US11/602,524
• Publication Number:
  US 20070124134A1
• Time in Patent Office: 1,736 Days
• Field of Search: 704/245, 704/251, 704/275, 379/88.01
• US Class Current: 704/275
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/197 Probabilistic grammars, e.g. word n-grams
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
• CPC Class Codes: G10L 2015/228 of application context
• CPC Class Codes: H04M 2250/74 with voice recognition means voice activated dialling H04M1/271; voice recognition algorithms G10L15/00

Abstract

Method for building a multimodal business channel between users, service providers and network operators. The service provided to the users is personalized with a user'"'"'s profile derived from language and speech models delivered by a speech recognition system. The language and speech models are synchronized with user dependent language models stored in a central platform made accessible to various value added service providers. They may also be copied into various devices of the user. Natural language processing algorithms may be used for extracting topics from user'"'"'s dialogues.

Claims

1. A method in a communication network for personalizing a service, comprising the steps of:

generating user dependant language models by a speech recognition system,

storing said user dependant language models,

making said user dependant language models, and/or a user profile derived from said user dependant language models, available to a software application running in a user'"'"'s device and/or available to external service providers, for personalizing an aspect of a service unrelated to speech processing,

using a microphone of a user end device of said user for gathering ambient speech material outside normal use of said user end device for voice or data communications with external devices, and

using said ambient speech material for adapting said user dependant language models.

2. The method of claim 1, wherein at least a subset of said user dependant language models is stored locally in a user device and synchronised with user dependant language models stored in a pervasive platform.

3. The method of claim 2, further comprising a step of performing speech recognition in said user device using said locally stored user dependant language models, wherein said aspect of a service is personalized using said centrally stored user dependant language models.

4. The method according to claim 2, further comprising the steps of:

loading into said user device an initial set of language models,

adapting said language models to said user,

synchronising said adapted, user dependant language models with said centrally stored user dependant language models, and

downloading said centrally stored user dependant language models into another device of said user.

5. The method according to claim 2, further comprising the steps of:

loading into said user device a set of language models that depends on the demographic and/or contextual data relating to said user, and/or on topics extracted from dialogues of said user,

adapting said language models in said user device, and

uploading said adapted language models into said platform.

6. The method according to claim 2, wherein previous versions of said user dependant language models are stored in said pervasive platform.

7. The method according to claim 1, further comprising the steps of:

gathering speech material related to the activities of the user,

performing speech recognition on said speech material, said speech recognition step comprising a step of adapting user dependant speech recognition language models,

retrieving from said user dependant speech recognition language models information used for adapting a user profile,

using said user profile for personalizing said service.

8. The method according to claim 7, comprising a step of making elements of said user profile and/or of said language models available in the Internet as Web services.

9. The method according to claim 1, further comprising a step of establishing a user dependant list of semantic expressions retrieved from speech material related to the activities of the user, and using said list for adapting said user profile.

10. The method according to claim 9, wherein said semantic expressions comprise at least one of the following:

words,

bigrams, trigrams and/or n-grams, and/or

semantic clusters and/or ontologies.

11. The method according to claim 9, further comprising a step of determining the frequency of use of said semantic expressions by said user and adapting said user profile according to said frequency.

12. The method according to claim 9, wherein said user profile depends on the time at which said semantic expressions have been gathered or spoken.

13. The method according to claim 9, wherein said user profile depends on the rareness of said semantic expressions and/or on the context when they were retrieved.

14. The method according to claim 1, further comprising the step of gathering in a remote speech server speech material spoken by a user when accessing said remote speech server.

15. The method according to claim 1, further comprising the step of gathering in a user end device speech material spoken by a user when speaking into said user end device.

16. The method according to claim 1, further comprising the steps of:

gathering speech material of said user during normal phone conversations with other users or devices, and

using said speech material for adapting said user dependant language models.

17. The method according to claim 1, further comprising the step of gathering non spoken semantic expressions entered on a keyboard or keypad by said user and/or displayed to said user, for example during multimodal interactions between said user and said device, and using said non spoken semantic expressions in addition to expressions recognised by said speech recognition system for said step of personalizing.

18. The method according to claim 17 wherein said non spoken semantic expressions include words and are at least partly gathered from Internet history logs, visited web pages, word processor applications, and/or e-mail client applications.

19. The method according to claim 17, wherein at least some of said non spoken semantic expressions include words that are retrieved from a personal word spelling dictionary.

20. The method according to claim 1, further comprising the step of making said user dependant language models available to a plurality of speech recognition systems in a communication network.

21. The method according to claim 1, wherein anonymized user profiles are made available to external value added service providers.

22. The method according to claim 1, wherein user profiles are only made available to external value added service providers authorised by said user.

23. The method according to claim 1, wherein said step of personalizing an aspect of a service includes sending of personalized advertising messages to said user.

24. The method according to claim 23, wherein said advertising messages are displayed on a multimodal reserved portion of a user'"'"'s device.

25. The method according to claim 24, wherein new messages are sent to said user device even outside communications initiated or answered by said user.

26. The method according to claim 23, further comprising the step of giving to said user an incentive for receiving said advertising messages.

27. The method according to claim 24, wherein said user may answer directly said advertising message by speaking and/or clicking and/or writing on said multimodal portion.

28. The method according to claim 27, further comprising a step of authenticating said user with his voice and/or with his fingerprint and/or with handwriting when he answers said advertising message.

29. The method according to claim 1, wherein said step of personalizing an aspect of a service includes proposing to said user names or addresses of users with matching interests or profiles.

30. The method according to claim 1, wherein said step of personalizing an aspect of a service includes selecting information matching the interests of said user.

31. The method according to claim 1, wherein said step of personalizing an aspect of a service includes ranking search results according to user criteria.

32. The method according to claim 1, wherein said step of personalizing an aspect of a service comprises filtering out spam messages.

33. The method according to claim 1, further comprising the step of using user dependant speech acoustic models delivered by said speech recognition system for personalizing said aspect of said service unrelated to speech processing.

34. A method used in a communication system for building a user profile with information about the interests and/or habits and/or preferences of the user, said method comprising the steps of:

generating user dependant language models by a speech recognition system,

deriving a user profile from said user dependant language models, and

providing a communication interface for transmitting at least a portion of said user profile to a provider computer.

35. A method used in a communication system for building a centrally stored user profile with information about the interests and/or habits and/or preferences of the user, said method comprising the steps of:

generating user dependant language models by a speech recognition system in a user device,

deriving a user profile from said user dependant language models,

synchronising said user dependant language models and/or user profiles with a central database of language models and/or user profiles remotely stored in a server, and

providing a communications interface for providing access to said language models and/or user profiles by external service provider computers for personalising services to the users.

36. A communication system in a communication network comprising:

a speech recognition system for generating a plurality of user dependant language models;

means for storing said user dependant language models; and

a communications interface for providing an external service provider and/or an application running in a user'"'"'s device access to said user dependant language models, and/or a user profile derived from said user dependant language models for personalizing an aspect of a service unrelated to speech processing.

37. A method in a communication network for personalizing a service, comprising the steps of:

generating user dependent language models by a speech recognition system,

storing said user dependent language models;

deriving from said user dependent language models a personalize list of semantic expressions; and

providing a communications interface for making at least a portion of said list available to an application running in a user'"'"'s device and/or available to external service providers for personalizing an aspect of a service unrelated to speech processing.

38. A communication system in a communication network comprising:

a speech recognition system for generating a plurality of user dependent language models;

means for storing said user dependent language models;

means for deriving from said user dependent language models a personalized list of semantic expressions; and

means a communications interface for providing an external service provider and/or an application running in a user'"'"'s device access to receive at least a portion of said list for personalizing an aspect of a service unrelated to speech processing.

39. A method in a communication network for personalizing a service, comprising the steps of:

generating user dependent language models by speech recognition system,

storing said user dependent language models in a platform,

providing an interface to said communication network for making said user dependent language models and/or a user profile comprising for each user a personalized list of semantic expressions derived from said user dependent language models available over a communication network to an external service provider, and

using said user dependent language models and/or user profile for personalizing a service provided by said external service provider.