SYSTEM AND METHOD FOR SPOKEN TOPIC OR CRITERION RECOGNITION IN DIGITAL MEDIA AND CONTEXTUAL ADVERTISING

• US 20090326947A1
• Filed: 06/26/2009
• Published: 12/31/2009
• Est. Priority Date: 06/27/2008
• Status: Abandoned Application

• Current Assignee: AdPassage, Inc.
• Original Assignee: AdPassage, Inc.
• Inventors: Arnold, James, Carter, P. Grant
• Application Number:
  US12/492,707
• Publication Number:
  US 20090326947A1
• Time in Patent Office: Days
• US Class Current: 704/257
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

Systems and methods for automated analysis and targeting of digital media based upon spoken topic or criterion recognition of the digital media are provided. Pre-specified criteria are used as the starting point for a top-down topic or criterion recognition approach. Individual words used in the audio track of the digital media are recognized only in context of each candidate topic or criterion hypothesis, thus yielding greater accuracy than two-step approaches that first transcribe speech and then recognize topic based upon the transcription.

Claims

1. A method of targeting one or more digital media for a spoken topic understanding application, comprising:

receiving one or more selection criteria;

performing a top-down criterion recognition of the digital media using the selection criteria as a starting point;

recognizing spoken words in the digital media in context of each selection criteria; and

identifying a first set of the digital media relevant to the selection criteria.

2. The method of claim 1, wherein performing the top-down criterion recognition does not include transcribing of the digital media.

3. The method of claim 1, wherein the spoken topic understanding application is an advertising application.

4. The method of claim 1, wherein the spoken topic understanding application is a non-advertising application.

5. The method of claim 1, wherein performing the top-down criterion recognition of the digital media comprises:

generating a broad criterion set from the selection criteria and pre-sorting the one or more digital media to the broad criterion set;

generating candidate criterion hypotheses at a finer granularity by using topically or demographically relevant query terms;

classifying the one or more digital media at the finer granularity.

6. The method of claim 5, wherein topically or demographically relevant query terms are obtained using metadata or inference on proprietary or publicly available ontologies.

7. The method of claim 1, further comprising training on digital media examples to generate one or more classification models for use in performing the top-down criterion recognition of the digital media.

8. The method of claim 7, further comprising based upon a particular application for the spoken topic understanding, calculating and incorporating a financial benefit of accurate identifications and a financial cost of inaccurate identifications into the classification models.

9. The method of claim 1, wherein selection criteria include one or more of a group consisting essentially of one or more topics, one or more names of products, one or more names of people, one or more places, items of commercial interest, and financial costs and benefits related to applications for spoken topic understanding, and further wherein performing top-down criterion recognition of the digital media comprises transforming the selection criteria into a set of search terms that distinguishes target categories and using a time-sampled probability function for each search term.

10. The method of claim 1, wherein selection criteria includes one or more of a group consisting essentially of targeted demographic, targeted viewer intent, one or more names of products, one or more names of people, one or more places, items of commercial interest, and financial costs and benefits related to applications for spoken topic understanding, and further wherein performing top-down criterion recognition of the digital media comprises transforming the selection criteria into a set of search terms that distinguishes demographic and viewer intent.

11. The method of claim 1, wherein performing the top-down criterion recognition of the digital media further comprises evaluating metadata associated with the digital media.

12. The method of claim 1, wherein performing the top-down criterion recognition of the digital media further comprises evaluating descriptive annotations associated with the digital media comprising on-line text descriptions, media source information, and information derived from other digital media processing technologies.

13. The method of claim 1, wherein performing the top-down criterion recognition of the digital media further comprises using computer speech recognition techniques and using natural language understanding techniques.

14. The method of claim 1, further comprising identifying a second set of the digital media for avoiding based upon a particular application for the spoken topic understanding.

15. A method of targeting one or more digital media for a spoken topic understanding advertising application, comprising:

receiving one or more advertising criteria;

generating a broad criterion set from the advertising criteria and pre-sorting the one or more digital media to the broad criterion set;

generating candidate criterion hypotheses at a finer granularity by using topically or demographically relevant query terms, wherein topically or demographically relevant query terms are obtained using metadata or inference on proprietary or publicly available ontologies;

classifying the one or more digital media at the finer granularity;

recognizing spoken words in the digital media in context of each advertising criteria; and

identifying a first set of the digital media for advertisement insertion.

16. The method of claim 15, further comprising identifying specific times within the first set of the digital media for advertisement placement.

17. The method of claim 15, further comprising integrating advertisement insertion information with advertisement servers.

18. The method of claim 15, further comprising integrating advertisement insertion information with advertising-serving platforms.

19. The method of claim 15, further comprising integrating advertisement insertion information with media buying consoles.

20. The method of claim 15, further comprising integrating advertisement insertion information with publisher advertisement management systems.

21. A system for targeting digital media based upon spoken criteria recognition of the digital media, comprising:

a communications module configured to receive one or more target criteria;

a model generation module configured to perform a top-down criterion recognition of the digital media using the target criteria as a starting point; and

an analyzer module configured to recognize spoken words in the digital media in context of each target criteria, wherein the system identifies a first set of the digital media relevant to the target criteria based upon the analysis.

22. The system of claim 21, further comprising a training database configured to store labeled digital media examples for training the system to generate classification models for use in performing the top-down criterion recognition of the digital media.

23. The system of claim 21 wherein the analyzer module does not transcribe one or more audio tracks associated with the digital media.

24. The system of claim 21, wherein performing the top-down criterion recognition of the digital media comprises:

generating a broad criterion set from the target criteria and pre-sorting the one or more digital media to the broad criterion set;

generating candidate criterion hypotheses at finer granularity by using topically or demographically relevant query terms;

classifying the one or more digital media at finer granularity.

25. The system of claim 21, further comprising a user profile database for storing information about user behavior and preferences.

26. The system of claim 21, further comprising one or more sources of digital media.

27. The system of claim 21, further comprising a media-management database for storing indices to particular ones of the digital media satisfying the target criteria.