Speech Recognition Language Models

• US 20110161081A1
• Filed: 12/22/2010
• Published: 06/30/2011
• Est. Priority Date: 12/23/2009
• Status: Abandoned Application

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Schalkwyk, Johan, Cohen, Michael H., Ballinger, Brandon M., Allauzen, Cyril Georges Luc
• Application Number:
  US12/977,017
• Publication Number:
  US 20110161081A1
• Time in Patent Office: Days
• US Class Current: 704/245
• CPC Class Codes: G06F 3/04886 by partitioning the display area of the touch-screen or the surface of the digitising tablet into independently controllable areas, e.g. virtual keyboards or menus
• CPC Class Codes: G06F 3/167 Audio in a user interface, e.g. using voice commands for navigating, audio feedback
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/58 Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
• CPC Class Codes: G10L 15/005 Language recognition
• 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 15/197 Probabilistic grammars, e.g. word n-grams
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Methods, computer program products and systems are described for forming a speech recognition language model. Multiple query-website relationships are determined by identifying websites that are determined to be relevant to queries using one or more search engines. Clusters are identified in the query-website relationships by connecting common queries and connecting common websites. A speech recognition language model is created for a particular website based on at least one of analyzing at queries in a cluster that includes the website or analyzing webpage content of web pages in the cluster that includes the website.

Claims

1. A computer-implemented method of forming a speech recognition language model, the method comprising:

determining, with a computer system, a plurality of query-website relationships by identifying websites that are determined to be relevant to queries using one or more search engines;

identifying clusters in the plurality of query-website relationships by connecting common queries and connecting common websites; and

creating a speech recognition language model for a particular website based on at least one of analyzing at queries in a cluster that includes the website or analyzing webpage content of web pages in the cluster that includes the website.

2. The method of claim 1, wherein determining the plurality of query-website relationships comprises analyzing search results returned by the one or more search engines to the queries.

3. The method of claim 1, wherein identifying clusters in the plurality of query-website relationships comprises analyzing a graph comprised of queries and websites at nodes of the graph, that are connected by edges defined by particular websites that are determined to be relevant to particular queries.

4. The method of claim 1, wherein a website is determined to be relevant to a query if the website is a top n ranked search result for the query in a set of ranked search results relevant to the query, wherein n is a predetermined integer.

5. The method of claim 1, wherein creating the speech recognition model for the particular website comprises creating cluster-based speech recognition models for each of a plurality of the clusters, and creating the speech recognition model by interpolating among cluster-based speech recognition models for clusters that surround the particular website.

6. A system for forming speech recognition language models comprising:

a data processing apparatus; and

storage coupled to the data processing apparatus storing code that when executed by the data processing apparatus causes the data processing apparatus to perform operations comprising;

determining a plurality of query-website relationships by identifying websites that are determined to be relevant to queries using one or more search engines;

identifying clusters in the plurality of query-website relationships by connecting common queries and common websites; and

creating a speech recognition language model for a particular website by at least one of analyzing queries in a cluster that holds the website or analyzing webpage content of web pages in the cluster that includes the website.

7. The system of claim 6, wherein determining the plurality of query-website relationships comprises analyzing search results returned by the one or more search engines to the queries.

8. The system of claim 6, wherein identifying clusters in the plurality of query-website relationships comprises analyzing a graph comprised of queries and websites at nodes of the graph, that are connected by edges defined by particular websites that are determined to be relevant to particular queries.

9. The system of claim 6, wherein a website is determined to be relevant to a query if the website is a top n ranked search result for the query in a set of ranked search results relevant to the query, wherein n is a predetermined integer.

10. The system of claim 6, wherein creating the speech recognition model for the particular website comprises creating cluster-based speech recognition models for each of a plurality of the clusters, and creating the speech recognition model by interpolating among cluster-based speech recognition models for clusters that surround the particular website.

11. A computer-readable storage device encoded with a computer program product, the computer program product including instructions for forming a speech recognition language model that, when executed, cause data processing apparatus to perform operations comprising:

determining a plurality of query-website relationships by identifying websites that are determined to be relevant to queries using one or more search engines;

identifying clusters in the plurality of query-website relationships by connecting common queries and connecting common websites; and

creating a speech recognition language model for a particular website based on at least one of analyzing queries in a cluster that includes the website or analyzing webpage content of web pages in the cluster that includes the website.

12. The computer-readable storage device of claim 11, wherein determining the plurality of query-website relationships comprises analyzing search results returned by the one or more search engines to the queries.

13. The computer-readable storage device of claim 11, wherein identifying clusters in the plurality of query-website relationships comprises analyzing a graph comprised of queries and websites at nodes of the graph, that are connected by edges defined by particular websites that are determined to be relevant to particular queries.

14. The computer-readable storage device of claim 11, wherein a website is determined to be relevant to a query if the website is a top n ranked search result for the query in a set of ranked search results relevant to the query, wherein n is a predetermined integer.

15. The computer-readable storage device of claim 11, wherein creating the speech recognition model for the particular website comprises creating cluster-based speech recognition models for each of a plurality of the clusters, and creating the speech recognition model by interpolating among cluster-based speech recognition models for clusters that surround the particular website.