Method and device for ranking video embeds

1. A method of preparing to deliver relevant video content responsive to search queries seeking video content, including:

collecting embedded video objects from RSS feeds and video hosting websites;

identifying duplicate content among the embedded video objects;

collecting and aggregating popularity statistics for at least some of the duplicate video objects from a plurality of the video hosting websites, wherein the popularity statistics include at least number of views;

parsing tags used to embed the video objects and text associated with the video objects in the RSS feeds and on the video hosting websites and producing, from the parsing, a plurality of terms associated with the video objects;

indexing on a machine readable storage medium the video objects for retrieval responsive to queries using the terms; and

ranking the video objects based on a combination of at least

a normalized number of views of a particular video object at particular video hosting websites based at least in part on an aggregate number of views at the particular video hosting websites and

expected quality metrics, including

quality experience with a source of the video object,

quality experience with the video hosting website where the video object resides, if different from the source, and

a likelihood that a link to the video object is not broken.

2. The method of claim 1 , further including:

collecting from the video hosting websites at least some comments on the video object attributed to viewers; and

including in the ranking combination a popularity statistic based on analysis of comments on the video object.

3. The method of claim 2 , further including in the analysis of the comments lexical analysis of the comments to exclude machine generated comments.

4. The method of claim 2 , further including in the analysis of the comments parsing the comments to find the terms to add to the indexing.

5. The method of claim 2 , further including in the analysis of the comments parsing the comments to evaluate the postiveness or negativeness of the comments.

6. The method of claim 2 , further including in the analysis of the comments grammatical analysis of the comments to evaluate the complexity of the comments.

7. The method of claim 2 , further including:

collecting from the video hosting websites counts of sharing the video object; and

including in the ranking combination a popularity statistic based on the counts of sharing the video object.

8. The method of claim 2 , further including combining the normalized number of views and the expected quality metrics as:

R

=

Views

MedianViews

×

(

1

-

PBroken

)

×

DSource

×

DHost

×

DComments

wherein R is a ranking score

Views/Median Views is the normalized number of views

PBroken is an estimated probability that a link to the embedded video object is broken

DSource is the quality experience with the source

DHost is the quality experience with the video hosting website

DComments is the popularity statistic based on the analysis of the comments.

9. The method of claim 1 , further including:

repeatedly collecting instances of the popularity statistics from the video hosting websites for a particular video object;

calculating one or more popularity velocity statistics based on at least two of the instances and time between collecting the instances; and

including in the ranking combination the popularity velocity statistics.

10. The method of claim 9 , further including combining the normalized number of views and the expected quality metrics as:

R

=

Views

MedianViews

×

(

1

-

PBroken

)

×

DSource

×

DHost

×

DUserRating

wherein R is a ranking score

Views/Median Views is the normalized number of views

PBroken is an estimated probability that a link to the embedded video object is broken

DSource is the quality experience with the source

DHost is the quality experience with the video hosting website

DUserRating is the viewer ratings.

11. The method of claim 1 , further including:

collecting from the video hosting websites viewer ratings of the video object; and

including in the ranking combination a popularity statistic based on the viewer ratings of the video object.

12. The method of claim 11 , further including combining the normalized number of views and the expected quality metrics as:

R

=

Views

MedianViews

×

(

1

-

PBroken

)

×

DSource

×

DHost

×

DSharing

wherein R is a ranking score

Views/Median Views is the normalized number of views

PBroken is an estimated probability that a link to the embedded video object is broken

DSource is the quality experience with the source

DHost is the quality experience with the video hosting website

DSharing is the popularity statistic based on the counts of sharing.

13. The method of claim 1 , further including:

collecting from the video hosting websites counts of sharing the video object; and

including in the ranking combination a popularity statistic based on the counts of sharing the video object.

14. The method of claim 1 , further including:

collecting from the video hosting websites counts of favoriting the video object; and

including in the ranking combination a popularity statistic based on the counts of favoriting the video object.

15. The method of claim 1 , further including:

collecting from the video hosting websites counts of adding the video object to a playlist; and

including in the ranking combination a popularity statistic based on the counts of adding the video object to a playlist.

16. The method of claim 1 , further including:

collecting from the video hosting websites counts of flagging the video object; and

including in the ranking combination a popularity statistic based on the counts of flagging the video object.

17. The method of claim 1 , wherein Views/Median Views, PBroken, and DSource are expressed in range(s) k(0.0 to 1.0) or a logarithmic equivalent, where k is a scaling factor or vector.