System and method for data-driven ad-insertion in online video streams

An apparatus comprising:

• one or more memories comprising computer-readable code, and one or more processors;
• wherein the one or more processors are configured, in response to execution of the computer-readable code, to cause the apparatus to at least:
• determine for a specific video an integer number k of locations for advertisements when streaming the specific video to a specific user that requests the specific video;
• select a set of advertisements to deliver with the specific video;
• utilize user feedback to determine an integer number k discrete locations within the specific video at which to place advertisements of the set;
• and
• electronically or optically deliver to the specific user the specific video with the set of advertisements dispersed among the k discrete locations for rendering on a local display device;
• wherein utilizing the user feedback to determine the k discrete locations comprises:
• storing a database indicating user preferences for various locations in videos for placing advertisements;
• and
• selecting from the database only those user preferences for a group of videos having a duration within a predetermined range that matches a duration of the selected video;
• and wherein the user preferences for the group of videos comprises a rating for each of multiple time-step locations t, and utilizing the user feedback to determine the k discrete locations further comprises:
• partitioning data in the database for the group of videos into an integer number P partitions such that each partition has a similar number of ratings;
• for each of the time-step locations t, forming a matrix R of the corresponding ratings;
• statistically fitting a line joining the time-step locations t with the matrices R;
• extracting from the line the k highest ratings, each highest rating spaced in time from one another by a minimum granularty;
• and
• determining the k discrete locations as the locations associated by the line with the k highest ratings.

A non-transitory computer readable memory tangibly storing program code which, when executed by a computing system, causes the computing system to at least:

• determine for a specific video an integer number k of locations for advertisements when streaming the specific video to a specific user that requests the specific video;
• select a set of advertisements to deliver with the specific video;
• utilize user feedback to determine an integer number k discrete locations within the specific video at which to place advertisements of the set;
• and
• electronically or optically deliver to the specific user the specific video with the set of advertisements dispersed among the k discrete locations for rendering on a local display device;
• wherein utilizing the user feedback to determine the k discrete locations comprises:
• storing a database indicating user preferences for various locations in videos for placing advertisements;
• and
• selecting from the database only those user preferences for a group of videos having a duration within a predetermined range that matches a duration of the selected video;
• and wherein the user preferences for the group of videos comprises a rating for each of multiple time-step locations t, and utilizing the user feedback to determine the k discrete locations further comprises:
• partitioning data in the database for the group of videos into an integer number P partitions such that each partition has a similar number of ratings;
• for each of the time-step locations t, forming a matrix R of the corresponding ratings;
• statistically fitting a line joining the time-step locations t with the matrices R;
• extracting from the line the k highest ratings, each highest rating spaced in time from one another by a minimum granularity;
• and
• determining the k discrete locations as the locations associated by the line with the k highest ratings.