Generation and maintenance of language model

• US 9,620,111 B1
• Filed: 05/01/2012
• Issued: 04/11/2017
• Est. Priority Date: 05/01/2012
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Bell, Matthew P., Polansky, Stephen, Watanabe, Yuzo
• Primary Examiner(s): Roberts, Shaun
• Application Number:
  US13/461,471
• Time in Patent Office: 1,806 Days
• Field of Search: 704231, 704251
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G10L 15/18 using natural language modelling
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: G10L 2015/226 using non-speech characteristics

Abstract

Described herein are systems and methods for the generation and maintenance of language models. Language models are developed based at least in part on transaction data from one or more users. These transactions may include purchases and other interactions between one or more users and one or more online merchants. The language models may be associated with a particular user or group of users. Ongoing transactions may modify the language models. The language models may be used to recognize spoken input from the one or more users.

Claims

1. A system, comprising:

at least one memory storing computer-executable instructions; and

at least one processor configured to access the at least one memory and execute the computer-executable instructions to;

determine an item associated with a transaction between a merchant and a user device using non-speech based user input transaction data associated with the transaction;

determine a context associated with the transaction using at least a portion of the transaction data, the context comprising a set of one or more language model elements;

determine, using the context, a first language model element of the set of one or more language model elements having a predefined association with the item, wherein the first language model element represents a command word or phrase;

modify a language model by including the first language model element in the language model, resulting in a modified language model;

determine, using the modified language model, that speech received by a user device includes the command word or phrase; and

cause an operation corresponding to the command word or phrase to be initiated in connection with the item.

2. The system of claim 1, the transaction comprising a purchase of a good or service from the merchant.

3. The system of claim 1, wherein the set of one or more language model elements comprises one or more words, phrases, or phonemes associated with a good or service.

4. The system of claim 1, wherein the language model is based at least in part on a respective probability of occurrence in speech of each of the set of one or more language model elements.

5. The system of claim 1, wherein the item comprises consumable content, and wherein the at least one processor is configured to initiate the operation to enable, at least in part, consumption of the consumable content.

6. The system of claim 1, wherein the transaction is non-speech based.

7. The system of claim 1, wherein the at least one processor is further configured to execute the computer-executable instructions to determine a category for the item, and wherein the at least one processor is configured to determine at least one language model element based at least in part on a predefined association with the category for the item.

8. The system of claim 1, wherein the language model comprises a second language model element, and wherein the at least one processor is further configured to execute the computer-executable instructions to:

determine that the second language model element was added to the language model prior to the first language model element;

determine a first weight assigned to the second language model element;

determine a second weight that is greater than the first weight; and

assign the second weight to the first language model element.

9. The system of claim 1, wherein to determine the context, the at least one processor is further configured to access acquisition data indicative of selection of a good for purchase;

and

identify the context based at least on the selection of the good for purchase.

10. The system of claim 1, wherein to determine the context, the at least one processor is further configured to access location data indicative of selection of a delivery address;

and

identify the context based at least on the selection of the delivery address.

11. A system, comprising:

at least one memory storing computer-executable instructions; and

at least one processor configured to access the at least one memory and execute the computer-executable instructions to;

determine a first item associated with a purchase transaction between a merchant and a user device using transaction data associated with the purchase transaction, the purchase transaction comprising a purchase of a good or a service from the merchant;

generate a purchase recommendation for a second item using at least a portion of the transaction data, wherein the second item is associated with the first item;

send the purchase recommendation to the user device;

determine one or more language model elements having a predefined association with the first item, the one or more language model elements comprising at least one language model element representative of the second item associated with the first item; and

modify a language model by including the at least one language model element in the language model.

12. The system of claim 11, wherein the at least one processor is further configured to modify the language model by executing the computer-executable instructions to remove a particular language model element from the language model or alter a probability of occurrence in human speech of the particular language model element.

13. The system of claim 11, wherein the language model is associated with the user or a group of users.

14. The system of claim 11, wherein to generate the purchase recommendation, the at least one processor is further configured to determine an association between third items and the item;

and

select the second item from the third items.

15. One or more non-transitory computer-readable media storing computer-executable instructions that, when executed by at least one processor, configure the at least one processor to perform operations comprising:

determining an item associated with a transaction between a merchant and a user device using non-speech based user input transaction data associated with the transaction;

determining a context associated with the transaction using at least a portion of the transaction data, the context comprising a set of one or more language model elements;

determining, using the context, a first language model element of the set of one or more language model elements having a predetermined association with the item, wherein the first language model element represents a command word or phrase;

modifying a language model by including the first language model element in the language model, resulting in a modified language model;

recognizing, using the modified language model, presence of the command word or phrase in input speech data; and

causing an operation corresponding to the command word or phrase to be initiated in connection with the item.

16. The computer-readable media of claim 15, the transaction comprising a purchase of the item from an online merchant.

17. The computer-readable media of claim 15, the operations further comprising:

identifying the user; and

selecting the language model from one or more candidate language models associated with the user.

18. The computer-readable media of claim 15, the operations further comprising adding a good or service to an order with the merchant.

19. The computer-readable media of claim 15, the language model comprising a plurality of language model elements.

20. The computer-readable media of claim 19, the plurality of language model elements based on one or more previous transactions.

21. One or more non-transitory computer-readable media storing computer-executable instructions that, when executed by at least one processor, configure the at least one processor to perform operations comprising:

determining an item associated with a transaction linked to a first user account;

determining a second user account associated with a recipient of the item;

determining a language model, wherein the language model comprises one or more language model elements having a predetermined association with the item associated with the transaction; and

associating the language model with the second user account, wherein operations of the item can be initiated with speech input corresponding to at least one of the one or more language model elements.

22. The computer-readable media of claim 21, wherein determining the second user account comprises determining that the second user account is associated with a delivery address specified in connection with the transaction.

23. The computer-readable media of claim 21, the transaction comprising one or more of a search, a purchase, an information request, or a location.

24. The computer-readable media of claim 21, wherein the item is a device, the operations further comprising providing at least a portion of the language model to the device for use in recognizing speech.