Video quality measures
A query including one or more keywords is received from a client device, and a content item (e.g., an advertisement) relevant to the one or more keywords is identified. A video is selected from one or more videos associated with the content item, the selection based on the one or more keywords. Display data is generated for displaying the content item at the client device, and video selection data is generated for displaying the video proximate to the content item at the client device. The display data and the video selection data are provided to the client device.
1. A computer-implemented method, comprising:
receiving, from a client device, a query including one or more keywords;
identifying a content item from a content item provider as being relevant to the one or more keywords;
accessing association data, specified by the content item provider, describing an association of a plurality of videos with the content item, the association data specifying each of the plurality of videos as being related to the content item, and further specifying that each of the plurality of videos may be displayed with the content item, wherein each of the plurality of videos is associated with metadata specific to the video;
for each video of the plurality of videos, calculating a score for the video, wherein the calculating comprises:
calculating a relevancy measure for the video based upon the one or more keywords matching metadata associated with the video; and
increasing the relevancy measure for each keyword match to the metadata by an amount proportionate to a feature weight of the matched metadata;
selecting a video from the plurality of videos associated with the content item based on the respective scores for the videos;
generating display data for displaying the content item at the client device;
generating video selection data for displaying the video proximate to the content item at the client device; and
providing the display data and the video selection data to the client device.
2. The method of claim 1 , wherein the video selection data causes the device to display a video element and causes the device to display the video associated with the content item proximate to the content item in response to a selection of the video element.
3. The method of claim 1 , wherein calculating a score for the video further comprising:
evaluating one or more signals associated with the video, the signals being representative of past user interactions in response to the video; and
calculating a quality score for the video based on the one or more signals.
4. The method of claim 3 , wherein the one or more signals include one or more of a play percentage; a pause percentage; a hidden percentage; a muted percentage; a conversion rate; and/or an information display rate.
5. The method of claim 1 , further comprising:
receiving the metadata from a content provider associated with each of the plurality of videos.
6. The method of claim 1 , wherein calculating a relevancy measure for the video based upon the one or more keywords matching metadata associated with the video comprises:
extracting audio from each of the videos;
comparing the audio to the one or more keywords; and
calculating the relevancy measure of the video based on a matching of the audio to the one or more keywords.
7. The method of claim 1 , wherein metadata specific to a corresponding video specify a subject matter of the corresponding video.
8. A computer-implemented method, comprising:
receiving, from a client device, a query including one or more keywords;
identifying a content item from a content item provider as being relevant to the one or more keywords;
accessing association data, specified by the content item provider, describing an association of a plurality of videos with the content item, the association specifying each of the plurality of videos as being related to the content item, and further specifying that each of the plurality of videos may be displayed with the content item, wherein each of the plurality of videos is associated with metadata specific to the video;
for each video of the plurality of videos:
calculating relevancy measures for the videos based on the one or more keywords and the metadata associated with the video and increasing the relevancy measures for each keyword match to the metadata by an amount proportionate to a feature weight of the matched metadata;
calculating a quality score for the video based upon signals associated with the video, the signals representative of past user interactions in response to the video;
calculating a final score for the video, the final score based on the product of the relevancy measure for the video and the quality score for the video; and
selecting a video based on the final scores.
9. The method of claim 8 , wherein the signals representative of past user interactions include one or more of a play percentage; a pause percentage; a hidden percentage; a muted percentage; a conversion rate; and/or an information display rate.
10. A system, comprising:
a data store configured to store advertisements and, for each advertisement, a corresponding plurality of videos associated with the advertisement, the association specifying each of the plurality of videos as being related to the advertisement, and further specifying that each of the plurality of videos may be displayed with the advertisement; and
an advertisement engine comprising instruction executable by a processing system and stored in a computer readable medium, wherein upon execution the instructions cause the processing system to perform the operations of:
receiving from a client device a query including one or more keywords;
identifying an advertisement from an advertiser relevant to the one or more keywords;
for each of the plurality of videos associated with the advertisement:
calculating a relevancy measure for the video based upon the one or more keywords matching metadata associated with the video;
calculating a score for the video based on a product of the relevancy measure and a quality score for the video;
selecting a video from the plurality of videos associated with the advertisement based on the respective scores for the videos, wherein the advertiser specifies the plurality of videos as being associated with the advertisement and that each of the plurality of videos may be displayed with the advertisement;
generating display data for displaying the advertisement at the client device;
generating video selection data for displaying the video proximate to the advertisement at the client device; and
provide the display data and the video selection data to the client device.
11. The system of claim 10 , wherein the video selection data causes the client device to display a video element and causes the client device to display the video associated with the advertisement proximate to the advertisement in response to a selection of the video element.
12. The system of claim 10 , wherein the the operation of calculating a score for the video comprises: evaluating one or more signals associated with the video, the signals being representative of past user interactions in response to the video; and
calculating the quality score for the video based on the one or more signals.
13. The system of claim 12 , wherein the signals associated with the video include one or more of a play percentage; a pause percentage; a hidden percentage; a muted percentage; a conversion rate; and/or an information display rate.
14. The system of claim 10 , wherein the operation of calculating a relevancy measure for the video based upon the one or more keywords matching metadata associated with the video comprises:
increasing the relevancy measure for each keyword match to the metadata by an amount proportionate to a feature weight of the matched metadata.
15. A system, comprising:
a data store configured to store advertisements and videos associated with the advertisements; and
an advertisement engine comprising instruction executable by a processing system and stored in a computer readable medium, wherein upon execution the instructions cause the processing system to perform the operations of:
receiving, from a client device, a query including one or more keywords from a client device;
identify an advertisement from an advertisement provider as being relevant to the one or more keywords;
accessing association data, specified by the advertisement provider, describing an association of a plurality of videos with the advertisement, the association specifying each of the plurality of videos as being related to the advertisement, and further specifying that each of the plurality of videos may be displayed with the advertisement;
for each of the plurality of videos:
calculating relevancy measures for the video based on the one or more keywords matching metadata associated with the video, wherein calculating the relevancy measures comprises increasing the relevancy measures for each keyword match to the metadata by an amount proportionate to a feature weight of the matched metadata;
calculating a quality score for the video based upon signals associated with the video, the signals representative of past user interactions in response to the video;
calculating a final score for the video, the final score based on the product of the relevancy measure for the video and the quality score for the video; and
selecting a video from the plurality of videos associated with the advertisement based on the final scores.
16. A system, comprising:
means for receiving a query including one or more keywords;
means for identifying a content item from a content item provider as being relevant to the one or more keywords, the content item associated with a plurality of videos by the content item provider, the association specifying each of the plurality of videos as being related to the content item and that may be displayed with the content item;
means for generating a relevancy measure for each of the videos based on the one or more keywords matching metadata associated with the video;
means for increasing the relevancy measures for each keyword match to the metadata by an amount proportionate to a feature weight of the matched metadata;
means for generating a quality score for each of the videos;
means for generating a final score for each of the videos, each final score based on a product of a respective relevancy measure and quality score for a video;
means for selecting a video from the plurality of videos based on final scores;
means for generating display data for displaying the content item at a client device;
means for generating video selection data for displaying the video proximate to the content item at the client device; and
means for providing the display data and the video selection data to the client device.
17. Software stored in a computer memory device accessible by a data processing apparatus that upon such execution cause the data processing apparatus to perform operations comprising:
receiving, from a client device, a query including one or more keywords;
identifying a content item from a content item provider as being relevant to the one or more keywords;
accessing association data, specified by the content item provider, describing an association of a plurality of videos with the content item, the association data specifying each of the plurality of videos as being related to the content item, and further specifying that each of the plurality of videos may be displayed with the content item, wherein each of the plurality of videos is associated with metadata specific to the video;
for each video of the plurality of videos, calculating a score for the video, wherein the calculating comprises:
calculating a relevancy measure for the video based upon the one or more keywords matching metadata associated with the video; and
increasing the relevancy measure for each keyword match to the metadata by an amount proportionate to a feature weight of the matched metadata;
selecting a video from the plurality of videos associated with the content item based on the respective scores for the videos;
generating display data for displaying the content item at the client device;
generating video selection data for displaying the video proximate to the content item at the client device; and
providing the display data and the video selection data to the client device.