Podcasts in personalized content streams

1. A method, comprising operations of:

obtaining a resource associated with a podcast from a first website publishing the podcast;

storing the resource on a second website that ingests content;

adding the resource as a leaf node to a taxonomy generated by the second website, wherein the addition is based at least in part on data associated with the podcast and a similarity measure that is cosine similarity or Jaccard similarity and wherein non-leaf nodes in the taxonomy are categories of content;

determining that a user of the second website is qualified as to at least one category that includes the resource as a leaf node, wherein the determination is based at least in part on feedback from the user that includes one or more of a viewing history for the user or a listening history for the user; and

serving the resource to the user in a content stream published by the second website, based at least in part on a personalization score associated with the resource, wherein each of the operations are executed by one or more processors.

2. The method of claim 1 , wherein the resource is an audio or video file and the podcast is a paid podcast.

3. The method of claim 1 , wherein the resource is a uniform resource locator (URL) for a web feed and the podcast is a free podcast.

4. The method of claim 1 , wherein the determination that the user of the second website is qualified as to at least one category is based at least in part on implicit relevance feedback from the user.

5. The method of claim 1 , wherein the data associated with the podcast includes data associated with the source of the podcast.

6. The method of claim 1 , wherein the data associated with the podcast includes data or metadata from the web pages of the first website.

7. The method of claim 1 , wherein addition of the resource as a leaf to the taxonomy includes analyzing an audio or video file associated with the resource.

8. The method of claim 1 , wherein the taxonomy is generated by the second website using supervised classifiers.

9. The method of claim 1 , wherein the personalization score is based at least in part on a parametric function that includes a measure of user fatigue as a parameter.

10. The method of claim 1 , wherein the personalization score results at least in part from an analysis of revenue.

11. One or more computer-readable media that are non-transitory and store instructions that, when executed by a processor, perform the following operations:

obtain a resource associated with a podcast from a first website publishing the podcast;

store the resource on a second website that ingests content;

add the resource as a leaf node to a taxonomy generated by the second website, wherein the addition is based at least in part on data associated with the podcast and a similarity measure re that is cosine similarity or Jaccard similarity and wherein non-leaf nodes in the taxonomy are categories of content;

determine that a user of the second website is qualified as to at least one category that includes the resource as a leaf node, wherein the determination is based at least in part on feedback from the user that includes one or more of a viewing history for the user or a listening history for the user; and

serve the resource to the user in a content stream published by the second website, based at least in part on a personalization score associated with the resource.

12. The computer-readable media of claim 11 , wherein the resource is an audio or video file and the podcast is a paid podcast.

13. The computer-readable media of claim 11 , wherein the resource is a uniform resource locator (URL) for a web feed and the podcast is a free podcast.

14. The computer-readable media of claim 11 , wherein the determination that the user of the second website is qualified as to at least one category is based at least in part on implicit relevance feedback from the user.

15. The computer-readable media of claim 11 , wherein the data associated with the podcast includes data associated with the source of the podcast.

16. The computer-readable media of claim 11 , wherein the data associated with the podcast includes data or metadata from the web pages of the first website.

17. The computer-readable media of claim 11 , wherein addition of the resource as a leaf to the taxonomy includes analyzing an audio or video file associated with the resource.

18. The computer-readable media of claim 11 , wherein the taxonomy is generated by the second website using supervised classifiers.

19. The computer-readable media of claim 11 , wherein the personalization score is based at least in part on a parametric function that includes a measure of user fatigue as a parameter.

20. A method, comprising operations of:

obtaining a resource associated with a podcast from a first website publishing the podcast;

storing the resource on a second website that ingests content;

adding the resource as a leaf node to a taxonomy generated by the second website, wherein the addition is based at least in part on data associated with the podcast and a similarity measure that is cosine similarity or Jaccard similarity, wherein non-leaf nodes in the taxonomy are categories of content based at least in part on data associated with the podcast, wherein the taxonomy can be represented as a B+ tree, and wherein the taxonomy is generated by the second website using supervised classifiers;

determining that a user of the second website is qualified as to at least one category that includes the resource as a leaf node, wherein the determination is based at least in part on feedback from the user that includes one or more of a viewing history for the user or a listening history for the user; and

serving the resource to the user in a content stream published by the second website, based at least in part on a personalization score associated with the resource, wherein each of the operations are executed by one or more processors.