Patent Application

App. No. 15/261,929

Abstract

Claims (76)

1 . A computer-implemented method for generating recommendation images related to a first audio media asset, comprising the steps of:

providing access to a computer memory configured to store a plurality of media assets;

providing access to a network, wherein the computer memory is connected to the network;

identifying at least a second audio media asset;

determining a first metadata document set related to the first audio media asset and a second metadata document set related to a second audio media asset;

calculating a first score vector of the first metadata document and a second score vector of the second metadata document;

determining whether the first audio media asset is similar to the second audio media asset;

in response to determining the first audio media asset is similar to the second audio media asset, constructing a common metadata document from the first metadata document and the second metadata document, wherein the common metadata document comprises the common terms of the first metadata document and the second metadata document;

identifying an image for each common term in the common metadata document;

adding the image and common term to a term-image pair set;

calculating a term-image score for each image and common term;

adding the term-image score to a term-image score set.

2 . The method of claim 1 , wherein calculating the first score vector and the second score vector comprises the step of using a weight function.

3 . The method of claim 2 , wherein the weight function is defined as ω (t, D), where t is a term, D is a document of terms, and co is a weight given to each term.

4 . The method of claim 3 , wherein the weight given to each term is an input of a user.

5 . The method of claim 3 , wherein the weight given to each term is automatically calculated.

6 . The method of claim 3 , wherein the weight given to each term is randomly generated.

7 . The method of claim 1 , wherein determining whether the first audio media asset is similar to the second audio media asset comprises the step of using a cosine similarity function.

8 . The method of claim 7 , wherein the cosine similarity function is defined as

σ



(

S

a

→

,

S

b

→

)

=

S

a

→

·

S

b

→



S

a

→







S

b

→



,

wherein {right arrow over (S a )} is the first score vector and {right arrow over (S b )} the second score vector.

9 . The method of claim 8 , wherein {right arrow over (S a )}= ω(t, D a )|t∈F c and {right arrow over (S b )}= ω(t, D b )|t∈F c , where t is a term, D is a document of terms, F c is the intersection of F a and F b , F a and F b and F c are features sets, and ω is a weight given to each term.

10 . The method of claim 1 , wherein the step of calculating a term-image score comprises the step of calculating the term-image score using P D ={(a, b)|a∈F D b∈I D pair (a, b) is tagged in document D}, where D is a document of terms, P D is the set of all terms tagged in document D, I D is the set of images against which the terms in P D are tagged in document D, and (a, b) the set of term-image pairs is tagged in document D.

11 . The method of claim 1 , further comprising the steps of:

determining a set of remaining metadata document, wherein each term in the set of remaining metadata document is in the first metadata document, and not in the common metadata document;

identifying a set of images of each term in the remaining metadata document;

calculating a term-image score for each image in the set of image of each term in the remaining metadata document; and

adding each term-image score to the term-image score set.

12 . The method of claim 1 , further comprising combining at least one image from the term-image pair set with the first audio media asset.

13 . The method of claim 12 , further comprising generating a video media asset from the combined images from the term-image pair set and the first audio media asset.

14 . The method of claim 1 , wherein at least one of the images of the term-image pair set comprises a visual effect.

15 . The method of claim 1 , wherein at least one of the images of the term-image pair set comprises textual content.

16 . The method of claim 1 , wherein at least the second audio media asset is stored in the computer memory.

17 . The method of claim 1 , wherein at least the second audio media asset is stored in a second computer memory.

18 . The method of claim 17 , wherein the second computer memory is connected to the network.

19 . The method of claim 17 , wherein the second computer memory is connected to a second network.

20 . A non-transitory computer-readable medium for generating recommendation images related to a first audio media asset, comprising instructions stored thereon, that when executed on a processor, perform the steps comprising:

identifying at least a second audio media asset;

determining a first metadata document set related to the first audio media asset and a second metadata document set related to a second audio media asset;

calculating a first score vector of the first metadata document and a second score vector of the second metadata document;

determining whether the first audio media asset is similar to the second audio media asset;

in response to determining the first audio media asset is similar to the second audio media asset, constructing a common metadata document from the first metadata document and the second metadata document, wherein the common metadata document comprises the common terms of the first metadata document and the second metadata document;

identifying an image for each common term in the common metadata document;

adding the image and common term to a term-image pair set;

calculating a term-image score for each image and common term;

adding the term-image score to a term-image score set.

Assignments (13)