Machine learning-based media content placement
View Patent ↗A system performs an automated analysis on a set of related media content items, such as static and video display advertisements for a coordinated advertising campaign. The analysis can include, for example, recognition of products, services, brands, objects, music, speech, motion, colors and moods, in order to determine content profiles for the content items. Different sequences of the media content items are placed within the web browsing paths of individual users, and the responses to the sequences are monitored with respect to desired outcomes, such as the purchase of a product or the visiting of an advertiser's website. The content profiles, the sequences, the placements, and the responses are provided as input into a machine learning system that is trained to select sequences and placements of media content items that achieve the desired outcomes. The system can be trained in part or wholly using feedback from its own output.
1. A method performed by a computer system having a processor and a memory, the method comprising, the computer system:
for each video segment of a plurality of video segments:
generating a content profile for the video segment by performing an automated analysis of the video segment, wherein the content profile comprises an inflection point specifying a time location within the video segment at which an event identified by the automated analysis occurs, and
for each user of a plurality of users to which the video segment is presented:
generating an interaction profile for the user with respect to the video segment based on user input to a computer in association with a presentation of the video segment to the user, wherein the interaction profile comprises an input time location within the video segment at which the user provided input, and
generating an interaction score based on the interaction profile and a comparison between the inflection point and the input time;
for a target video segment of the plurality of video segments, selecting a target plurality of users based on:
the content profile for the target video segment, and
for each of at least one subject video segment selected from the plurality of video segments:
the content profile for the subject video segment, and
the plurality of interaction scores for users with respect to the subject video segment; and
causing the target video segment to be presented to at least one of the selected target plurality of users.
2. The method of claim 1 , further comprising generating a user profile for each user of the plurality of users, wherein the selecting a target plurality of users is further based on the user profiles.
3. The method of claim 2 , wherein the user profile for each user comprises at least one item selected from a group consisting of: location of the user, geolocation of the user, age of the user, gender of the user, and personality of the user.
4. The method of claim 1 , wherein the content profile further comprises a plurality of attribute fields each of which attribute fields specifies a characteristic of the video segment determined based on the automated analysis, and wherein one of the attribute fields is configured to store a category assigned to the video segment by the automated analysis.
5. The method of claim 2 , wherein selecting the target plurality of users is further based on the plurality of interaction profiles for users with respect to the subject video segment.
6. The method of claim 5 , wherein selecting the target plurality of users is further based on the plurality of interaction profiles for users with respect to the target video segment.
7. The method of claim 6 , wherein the selecting a target plurality of users comprises:
determining an engagement likelihood score based on the content profile for the target video segment and further based on, for each of the at least one subject video segment selected from the plurality of video segments, the content profile for the subject video segment and the plurality of interaction profiles for users with respect to the subject video segment,
wherein the target plurality of users is selected based on the engagement likelihood score.
8. The method of claim 7 , wherein the user input for at least one user comprises active user input selected from the group consisting of: keyboard input, pointer selection, button press, voice capture, and touch screen selection.
9. The method of claim 7 , wherein the user input for at least one user comprises passive user input selected from the group consisting of: visibility of the presentation of the video segment on a computer display, absence of user input interrupting the presentation of the video segment, an indication that the presentation of the video segment continued to completion, and use of a camera to determine whether the user is viewing the presentation of the video segment.
10. The method of claim 7 , wherein the user input for at least one user comprises visibility of the video segment on a computer display determined based on whether another application window or scrolling of a page partly or fully obscures display of the video segment.
11. The method of claim 1 , wherein the interaction profile for at least one user comprises one or more interaction fields.
12. The method of claim 11 , wherein one of the at least one interaction fields indicates whether the video segment was presented to the user in its entirety.
13. The method of claim 11 , wherein one of the at least one interaction fields indicates how much of the video segment was presented to the user.
14. The method of claim 11 , wherein one of the at least one interaction fields indicates a time location within the video segment at which time location the user selected a selectable link presented in association with the video segment.
15. The method of claim 1 , wherein the interaction profile comprises an indication of intent by the user to take an action.
16. The method of claim 15 , wherein the indication of intent by the user to take an action is obtained in response to a survey presented to the user in association with the presentation of the video segment to the user.
17. The method of claim 1 , wherein the selecting a target plurality of users is performed using a trained machine learning system.
18. The method of claim 17 , wherein the trained machine learning system is trained using the content profiles for the plurality of video segments and the interaction profiles for the plurality of users.
19. The method of claim 18 , wherein the trained machine learning system is a neural network.
20. The method of claim 18 , wherein the trained machine learning system is a Bayesian network.