Systems and methods for automated content curation using signature analysis
Systems and methods are described herein for curating content that follows a narrative structure. A narrative structure comprises narrative portions that have a defined order. Signature analysis of known content that follows the narrative structure is used to train machine learning models for the narrative structure and the narrative portions that make up the narrative structure. Signature analysis of candidate content segments, along with machine learning models for the narrative portions, are used to identify candidate content segments that match the respective narrative portions. A candidate playlist is generated of the identified candidate content segments in the defined order. In one embodiment, the machine learning model for the narrative structure is used to validate the generated playlist.
1. A method for curating content, the method comprising:
generating, for each of a plurality of candidate segments of a content to be curated, a segment signature vector based on content of the candidate segment;
retrieving, for each portion of a plurality of portions of a narrative structure, a respective segment machine learning model, wherein the plurality of portions of the narrative structure have a defined order;
analyzing, for each portion of the narrative structure, the segment signature vector using a respective segment machine learning model to identify a segment signature vector match;
identifying, for each portion of the narrative structure, a candidate segment that corresponds to the identified segment signature vector match; and
generating a candidate playlist for the curated content, wherein the candidate playlist includes the identified candidate segments of the curated content in the defined order.
2. The method of claim 1 wherein the segment signature vectors are generated based on characteristics associated with audio of the content.
3. The method of claim 1 wherein the segment signature vectors are generated based on characteristics associated with video frames of the content.
4. The method of claim 1 further comprising:
generating a content signature vector based on content of the candidate playlist;
retrieving a content machine learning model; and
analyzing the content signature vector using the content machine learning model to determine whether there is a match.
5. The method of claim 4 further comprising in response to determining that there is a match, validating the candidate playlist.
6. The method of claim 5 further comprising distributing the validated candidate playlist for presentation on a user interface.
7. The method of claim 5 further comprising updating the segment machine learning model and the content machine learning model based on the validated candidate playlist.
8. The method of claim 4 further comprising in response to determining that there is no match, identifying one or more candidate segments of the candidate playlist that caused the content signature vector of the candidate playlist to not match.
9. The method of claim 8 further comprising replacing the identified one or more candidate segments with one or more new candidate segments.
10. The method of claim 4 wherein generating the content signature vector comprises applying mathematical operations across all of the identified candidate segments of the candidate playlist in the defined order.
11. A system comprising:
a memory; and
control circuitry configured to:
generate, for each of a plurality of candidate segments of a content to be curated, a segment signature vector based on content of the candidate segment;
retrieve, for each portion of a plurality of portions of a narrative structure, a segment machine learning model, wherein the plurality of portions of the narrative structure have a defined order;
analyze, for each portion of the narrative structure, the segment signature vector using a respective segment machine learning model to identify a segment signature vector match;
identify, for each portion of the narrative structure, a candidate segment that corresponds to the identified segment signature vector match; and
generate a candidate playlist for the curated content, wherein the candidate playlist includes the identified candidate segments of the curated content in the defined order.
12. The system of claim 11 wherein the segment signature vectors are generated based on characteristics associated with audio of the content.
13. The system of claim 11 wherein the segment signature vectors are generated based on characteristics associated with video frames of the content.
14. The system of claim 11 wherein the control circuitry is further configured to:
generate a content signature vector based on content of the candidate playlist;
retrieve a content machine learning model; and
analyze the content signature vector using the content machine learning model to determine whether there is a match.
15. The system of claim 14 wherein the control circuitry is further configured to:
validate the candidate playlist in response to determining that there is a match.
16. The system of claim 15 wherein the control circuitry is further configured to:
distribute the validated candidate playlist for presentation on a user interface.
17. The system of claim 15 wherein the control circuitry is further configured to:
update the segment machine learning model and the content machine learning model based on the validated candidate playlist.
18. The system of claim 14 wherein the control circuitry is further configured to:
identify one or more candidate segments of the candidate playlist that caused the content signature vector of the candidate playlist to not match in response to determining that there is no match.
19. The system of claim 18 wherein the control circuitry is further configured to:
replace the identified one or more candidate segments with one or more new candidate segments.
20. The system of claim 14 wherein to generate the content signature vector, the control circuitry is further configured to:
apply mathematical operations across all of the identified candidate segments of the candidate playlist in the defined order.