Dynamically determining the relatedness of web objects

1. A method comprising:

receiving, by a processor, a new web object, the new web object having properties that describe behaviors of response to a roll over, hover, and click, the web object being sizable, formattable, and manipulable, the web object being manageable using one of source code and auxiliary software, the web object comprising at least one of a video or an image;

identifying, by the processor, a first cluster of web objects from a first data structure comprising web objects that are organized into clusters based on a click-through relatedness;

identifying, by the processor, a second cluster of web objects from a second data structure comprising web objects that are organized into clusters based on a metadata relatedness; and

selecting, in real time, by the processor, an output set of web objects related to the new web object from the first and second clusters of web objects.

2. The method of claim 1 , wherein the web objects comprise at least one of a video, an image, or a document.

3. The method of claim 1 , further comprising:

generating clusters from a plurality of web objects, each cluster based on a relatedness of a set of web objects, the relatedness being a value representing frequency of selections and similarity of metadata terms, the similarity being quantified using a cosine similarity metric.

4. The method of claim 1 , further comprising:

adding the new web object to the first and second data structures during batch processing.

5. The method of claim 1 , further comprising:

identifying one or more clusters based on metadata associated with the new web object, the metadata having been tokenized and filtered; and

calculating a relatedness between the new web object and the one or more clusters using a weighted cosine distance, the relatedness comparing metadata of the new web object against metadata of the one or more clusters.

6. The method of claim 5 , wherein the one or more clusters includes the first set of web objects based on a click-through relatedness.

7. The method of claim 5 , wherein the one or more clusters includes the second set of web objects based on metadata relatedness.

8. The method of claim 5 , wherein the metadata comprises at least one of a search term or a description of the new web object.

9. A non-transitory computer-readable storage medium that when executed by a processor, performs a method comprising:

receiving, by the processor, a new web object, the new web object having properties that describe behaviors of response to a roll over, hover, and click, the web object being sizable, formattable, and manipulable, the web object being manageable using one of source code and auxiliary software, the web object comprising at least one of a video or an image;

identifying, by the processor, a first cluster of web objects from a first data structure comprising web objects that are organized into clusters based on a click-through relatedness;

identifying, by the processor, a second cluster of web objects from a second data structure comprising web objects that are organized into clusters based on a metadata relatedness; and

selecting, in real time, by the processor, an output set of web objects related to the new web object from the first and second clusters of web objects.

10. The medium of claim 9 , wherein the web objects comprises at least one of a video, an image, or a document.

11. The medium of claim 9 , further comprising:

generating clusters from a plurality of web objects, each cluster based on a relatedness of a set of web objects.

12. The medium of claim 9 , further comprising:

adding the web object to the first and second data structures during batch processing.

13. The medium of claim 9 , further comprising:

identifying one or more clusters based on metadata associated with the new web object, the metadata having been tokenized and filtered; and

calculating a relatedness between the new web object and the one or more clusters using a weighted cosine distance, the relatedness comparing metadata of the new web object against metadata of the one or more clusters.

14. The medium of claim 13 , wherein the one or more clusters includes the first set of web objects based on a click-through relatedness.

15. The medium of claim 13 , wherein the one or more clusters includes the second set of web objects based on a metadata relatedness.

16. The medium of claim 13 , wherein the metadata comprises at least one of a search term or a description of the new web object.

17. A system comprising:

a processor;

a memory;

a client communication module, executed by the processor and stored in the memory, to receive an indication of a new web object, the new web object having properties that describe behaviors of response to a roll over, hover, and click, the web object being sizable, formattable, and manipulable, the web object being manageable using one of source code and auxiliary software, the web object comprising at least one of a video or an image;

a cluster identifier, executed by the processor, stored in the memory, and communicatively coupled to the client communication module, the cluster identifier identifying a first cluster of web objects from a first data structure stored in a database and comprising web objects that are organized into clusters based on a click-through relatedness, the cluster identifier identifying a second cluster of web objects from a second data structure stored in the database and comprising web objects that are organized into clusters based on metadata relatedness; and

an object selector, executed by the processor, stored in the memory, and communicatively coupled to the cluster identifier, the object selector selecting in real time an output set of web objects related to the new web object from the first and second clusters of web objects.

18. The system of claim 17 :

wherein the cluster identifier identifies one or more clusters based on metadata associated with the new web object, the metadata having been tokenized and filtered; and

wherein a relatedness is calculated between the new web object and the one or more clusters using a weighted cosine distance, the relatedness comparing metadata of the new web object against metadata of the one or more clusters.