Ranking advertisement(s) based upon advertisement feature(s)

• US 8,620,912 B2
• Filed: 06/16/2010
• Issued: 12/31/2013
• Est. Priority Date: 06/16/2010
• Status: Active Grant

• Current Assignee: Zhigu Holdings Limited
• Original Assignee: Microsoft Corporation
• Inventors: Wang, Xin-Jing, Zhang, Lei, Ma, Wei-Ying
• Primary Examiner(s): LE, UYEN T
• Application Number:
  US12/816,533
• Publication Number:
  US 20110314013A1
• Time in Patent Office: 1,294 Days
• Field of Search: None
• US Class Current: 707/731
• CPC Class Codes: G06Q 30/0241 Advertisements
• CPC Class Codes: G06Q 30/0251 Targeted advertisements

Abstract

While browsing, a user may interact with a wide variety of images. The user may upload and share images taken with a digital camera and/or search for image using a search engine. Because images are rich in contextual information, it may be advantageous to provide additional information, such as adjacent market advertising based upon matching advertisements with contextual information of the images. Accordingly, a query image may be used to retrieve a video frame set. The video frame set may be expanded with related video frames corresponding to adjacent markets. The expanded video frame set may be grouped into clusters of similar frames. The clusters may be used to rank advertisements based upon how similar the advertisements are to the clusters and/or video frames within the clusters. In this way, one or more ranked advertisements may be presented with the query image.

Claims

1. A method for ranking an ad, comprising:

retrieving an initial video frame set comprising one or more initial video frames based upon a query image;

identifying one or more related video frames related to at least one initial video frame of the initial video frame set to create an expanded video frame set comprising one or more video frames, the one or more video frames of the expanded video frame set comprising at least some initial video frames and at least some related video frames, the expanded video frame set not comprising the query image;

grouping at least some of the one or more video frames of the expanded video frame set into one or more clusters; and

ranking an ad based upon an ad feature of the ad corresponding to a cluster feature of a cluster, at least some of at least one of the retrieving, the identifying, the grouping, or the ranking implemented at least in part via a processing unit.

2. The method of claim 1, the identifying one or more related video frames comprising:

identifying a related video frame comprising a first feature corresponding to a second feature of an initial video frame.

3. The method of claim 2, the query image not associated with the first feature.

4. The method of claim 1, the identifying one or more related video frames comprising:

identifying a related video frame based upon the related video frame occurring within a video comprising an initial video frame.

5. The method of claim 1, the grouping comprising:

defining a cluster based upon one or more features extracted from the expanded video frame set.

6. The method of claim 1, the grouping comprising:

executing a multimodal Dirichlet Process Mixture Sets model upon the expanded video frame set to define a cluster.

7. The method of claim 1, the retrieving an initial video frame set comprising:

obtaining an initial video frame that is visually similar to the query image.

8. The method of claim 1, the retrieving an initial video frame set comprising:

obtaining an initial video frame that is textually similar to the query image.

9. The method of claim 1, the retrieving an initial video frame set comprising:

retrieving an initial video frame from a large scale video frame database based upon matching a query image feature of the query image with a feature of the initial video frame.

10. The method of claim 1, comprising:

ranking one or more ads based upon the one or more clusters to create a set of ranked ads.

11. The method of claim 10, comprising:

presenting a ranked ad from the set of ranked ads based upon a rank of the ranked ad.

12. The method of claim 1, the ranking an ad comprising:

ranking one or more ad frames of the ad.

13. The method of claim 1, the ranking comprising:

executing a multimodal Dirichlet Process Mixture Sets model upon the ad using the one or more clusters.

14. A system for ranking an ad, comprising:

a frame set retriever configured to;

retrieve an initial video frame set comprising one or more initial video frames based upon a query image, the initial frame set not comprising the query image;

a frame set expander configured to;

identify one or more related video frames related to at least one initial video frame of the initial video frame set to create an expanded video frame set comprising one or more video frames, the one or more video frames of the expanded video frame set comprising at least some initial video frames and at least some related video frames;

a cluster generator configured to;

group at least some of the one or more video frames of the expanded video frame set into one or more clusters; and

an ad ranker configured to;

rank an ad based upon an ad feature of the ad corresponding to a cluster feature of a cluster, at least some of at least one of the frame set retriever, the frame set expander, the cluster generator, or the ad ranker implemented at least in part via a processing unit.

15. The system of claim 14, the frame set expander configured to:

identify a related video frame comprising a first feature corresponding to a second feature of an initial video frame, the query image not associated with the first feature.

16. The system of claim 14, the frame set expander configured to:

identify a related video frame based upon the related video frame occurring within a video comprising an initial video frame.

17. The system of claim 14, comprising:

a presentation component configured to;

present the ad based upon a rank assigned to the ad.

18. The system of claim 14, the cluster generator configured to:

execute a multimodal Dirichlet Process Mixture Sets model upon the expanded video frame set to define a cluster.

19. The system of claim 14, the ad ranker configured to:

execute a multimodal Dirichlet Process Mixture Sets model upon the ad using the one or more clusters.

20. A computer-readable device comprising processor-executable instructions that when executed perform a method for ranking an ad, comprising:

retrieving an initial video frame set comprising one or more initial video frames based upon a query image;

identifying one or more related video frames, not comprising the query image, related to at least one initial video frame of the initial video frame set to create an expanded video frame set comprising one or more video frames, the one or more video frames of the expanded video frame set comprising at least some initial video frames and at least some related video frames;

grouping at least some of the one or more video frames of the expanded video frame set into one or more clusters; and

ranking an ad based upon an ad feature of the ad corresponding to a cluster feature of a cluster.