Media file recommendations for a search engine

1 . A computer-implemented method, comprising:

receiving, in a search engine, a search query for a media file from a user;

identifying, using a neural network trained to represent a style preference of the user as a region of a multi-dimensional space, the style preference of the user encoded by one or more media file attributes of media files in a user-related search history;

selecting, from a database, a one or more media files based on the search query and the style preference of the user;

determining, according to a distance in the multi-dimensional space of a representation of each of the one or more media files from the region of the multi-dimensional space, a style preference score for the one or more media files, the representation of each of the one or more media files determined based on the one or more media file attributes; and

recommending to the user a top ranked media file based on the style preference score.

2 . The computer-implemented method of claim 1 , wherein the identifying of the style preference of the user comprises selecting the style preference from a cutoff portion of the user-related search history that is older than a pre-selected time.

3 . The computer-implemented method of claim 1 , wherein the identifying of the style preference of the user comprises identifying the one or more media file attributes from a metadata, a color selection, an angle of view, a texture, a visual pattern, a contrast, and a combination thereof.

4 . The computer-implemented method of claim 1 , wherein the identifying of the style preference of the user includes comparing the one or more media file attributes with salient attributes from the media files in the user-related search history.

5 . The computer-implemented method of claim 1 , wherein the identifying of the style preference of the user comprises selecting a media file that has been clicked on by the user according to the user-related search history.

6 . The computer-implemented method of claim 1 , wherein the identifying of the style preference of the user comprises identifying a geographic origin of the user from a metadata attribute of a respective one of the media files.

7 . The computer-implemented method of claim 1 , wherein the identifying of the style preference for of the user comprises identifying a professional association of the user.

8 . The computer-implemented method of claim 1 , further comprising verifying that the user-related search history includes one or more media files associated with a semantically different search query.

9 . The computer-implemented method of claim 1 , further comprising receiving, in the search engine from a content producer, a new media file, identifying a salient attribute in the new media file, updating a style preference for the content producer based on the salient attribute, and storing the new media file with the style preference for the content producer in a database.

10 . The computer-implemented method of claim 1 , further comprising defining a new style preference for the user based on the search query, and storing the new style preference in a database.

11 . A system, comprising:

a memory storing instructions; and

one or more hardware processors configured to execute the instructions to cause the system to:

receive, in a search engine, a search query for a media file from a user;

identify, using a neural network trained to represent a style preference of the user as a region of a multi-dimensional space, the style preference of the user encoded by one or more media file attributes of media files in a user-related search history;

select, from a database, a one or more media files based on the search query and the style preference of the user;

determine, according to a distance in the multi-dimensional space of a representation of each of the one or more media files from the region of the multi-dimensional space, a style preference score for the one or more media files, the representation of each of the one or more media files determined based on the one or more media file attributes; and

recommend to the user a top ranked media file based on the style preference score.

12 . The system of claim 11 , wherein to identify the style preference of the user, the instructions are executed to select the style preference from a cutoff portion of the user-related search history that is older than a pre-selected time.

13 . The system of claim 11 , wherein to identify the style preference of the user, the instructions are executed to identify the one or more media file attributes from a metadata, a color selection, an angle of view, a texture, a visual pattern, a contrast, and a combination thereof.

14 . The system of claim 11 , wherein to identify the style preference of the user, the instructions are executed to compare the one or more media file attributes with salient attributes from the media files in the user-related search history.

15 . The system of claim 11 , wherein to identify a style preference of the user, the instructions are executed to select a media file that has been clicked on by the user according to the user-related search history.

16 . A non-transitory computer-readable medium storing a program, which when executed by a computer, configures the computer to:

receive, in a search engine, a search query for a media file from a user;

identify, using a neural network trained to represent a style preference of the user as a region of a multi-dimensional space, the style preference of the user encoded by one or more media file attributes of media files in a user-related search history;

select, from a database, a one or more media files based on the search query and the style preference of the user;

determine, according to a distance in the multi-dimensional space of a representation of each of the one or more media files from the region of the multi-dimensional space, a style preference score for the one or more media files, the representation of each of the one or more media files determined based on the one or more media file attributes; and

recommend to the user a top ranked media file based on the style preference score.

17 . The non-transitory computer-readable medium of claim 16 , wherein to identify the style preference of the user, the program further configures the computer to select the style preference from a cutoff portion of the user-related search history that is older than a pre-selected time.

18 . The non-transitory computer-readable medium of claim 16 , wherein to identify the style preference of the user, the program further configures the computer to identify the one or more media files attributes from a metadata, a color selection, an angle of view, a texture, a visual pattern, a contrast, and a combination thereof.

19 . The non-transitory computer-readable medium of claim 16 , wherein to identify the style preference of the user, the program further configures the computer to compare the one or more media file attributes with salient attributes from the media files in the user-related search history.

20 . The non-transitory computer-readable medium of claim 16 , wherein to identify the style preference of the user, the program further configures the computer to select a media file that has been clicked on by the user according to the user-related search history.