Method for atomically tracking and storing video segments in multi-segment audio-video compositions

Abstract

A method includes accessing an audiovisual composition comprising a target video segment and a source video segment. The method also includes, in response to presence of the target video segment and the source video segment in the audiovisual composition: accessing a first keyword associated with the source video segment; and calculating a first relevance score for the first keyword relative to the target video segment based on a temporal position of the source video segment in the audiovisual composition and a temporal position of the target video segment in the audiovisual composition; accessing a textual query comprising the first keyword. The method additionally includes: generating a first query result based on the textual query, the first query result comprising the target video segment based on the first relevance score; and at a native composition application, rendering a representation of the first query result.

I claim: 
     
         1 . A method comprising:
 accessing an audiovisual composition comprising:
 a target video segment represented by a target video node; and 
 a source video segment represented by a source video node; 
   identifying a first keyword for the source video segment, the first keyword represented by a first keyword node;   calculating a primary relevance score for the first keyword relative to the source video segment;   updating a first edge connecting the first keyword node and the source video node based on the primary relevance score;   calculating a secondary relevance score for the first keyword relative to the target video segment based on:
 a temporal position of the source video segment in the audiovisual composition; 
 a temporal position of the target video segment in the audiovisual composition; and 
 the primary relevance score for the first keyword relative to the source video segment; and 
   updating a second edge connecting the first keyword node and the target video node based on the secondary relevance score.   
     
     
         2 . The method of  claim 1 , further comprising:
 accessing a textual query comprising the first keyword;   in response to the textual query comprising the first keyword:
 identifying the first keyword node representing the first associative keyword; 
 traversing the first edge from the first keyword node to access the source video node; 
 appending the source video segment represented by the source video node to a first query result based on the first edge; 
 traversing the second edge from the first keyword node to access the target video node; and 
 appending the target video segment represented by target video node to the first query result based on the second edge; and 
   rendering a representation of the first query result.   
     
     
         3 . The method of  claim 2 , wherein accessing the textual query comprising the first keyword comprises, in response to receiving a textual input comprising the first keyword at the native composition application, appending the first keyword to the textual query. 
     
     
         4 . The method of  claim 2 , wherein accessing the textual query comprising the first keyword comprises, in response to receiving a selection of a selected video segment for inclusion in a new audiovisual composition at the native composition application:
 identifying the first keyword based on the selected video segment; and   appending the first keyword to the textual query.   
     
     
         5 . The method of  claim 2 , wherein accessing the textual query comprising the first keyword comprises:
 in response to receiving a textual input comprising a second keyword at the native composition application, appending the second keyword to the textual query;   in response to receiving a selection of a selected video segment for inclusion in a new audiovisual composition at the native composition application:
 identifying a third keyword based on the selected video segment; and 
 appending the third keyword to the textual query; and 
   wherein one of a group consisting of the second keyword and the third keyword matches the first keyword.   
     
     
         6 . The method of  claim 1 , wherein identifying the first keyword for the source video segment comprises:
 executing a taxonomic video classification model on the source video segment to generate a classification vector associated with the source video segment; and   extracting the first keyword based on a classification in the classification vector.   
     
     
         7 . The method of  claim 6 , wherein calculating the primary relevance score for the first keyword relative to the source video segment comprises calculating the primary relevance score for the first descriptive keyword based on a confidence score for the classification in the classification vector. 
     
     
         8 . The method of  claim 1 :
 wherein accessing the audiovisual composition comprises accessing the audiovisual composition comprising:
 the target video segment; 
 the source video segment; and 
 audio content; and 
   wherein identifying the first keyword for the source video segment comprises:
 identifying a concurrent audio segment of the audio content to the source video segment; 
 executing a speech recognition model on the concurrent audio segment to generate a set of audible words; and 
 extracting the first keyword from the set of audible words. 
   
     
     
         9 . The method of  claim 1 , wherein identifying the first keyword for the source video segment comprises receiving a user input designating the first keyword. 
     
     
         10 . The method of  claim 1 , wherein identifying the first keyword for the source video segment comprises:
 accessing a timestamped comment concurrent with the source video segment; and   extracting the first keyword from the timestamped comment.   
     
     
         11 . The method of  claim 1 , wherein calculating the secondary relevance score for the first keyword relative to the target video segment comprises:
 calculating an adjacency of the source video segment to the target video segment based on the temporal position of the source video segment in the audiovisual composition and the temporal position of the target video segment in the audiovisual composition; and   calculating the secondary relevance score for the first keyword based on the adjacency of the source video segment to the target video segment and the primary relevance score for the first keyword relative to the source video segment.   
     
     
         12 . The method of  claim 1 :
 wherein accessing the audiovisual composition comprises accessing the audiovisual composition comprising the target video segment and the source video segment, the target video segment associated with a first user; and   further comprising:
 generating a report for the target video segment comprising a set of keywords for the target video segment, the set of keywords comprising the first keyword; and 
 serving the report to the first user. 
   
     
     
         13 . The method of  claim 12 , wherein generating the report for the target video segment comprises:
 calculating a sentiment score based on the set of keywords for the target video segment; and   in response to the sentiment score exceeding a threshold sentiment score:
 generating the report for the target video segment; and 
 serving the report to the first user. 
   
     
     
         14 . The method of  claim 1 :
 wherein updating the first edge comprises:
 accessing a first cumulative relevance score associated with the first edge; and 
 updating the first cumulative relevance score based on the primary relevance score; 
   wherein updating a second edge comprises:
 accessing a second cumulative relevance score associated with the second edge; and 
 updating the second cumulative relevance score based on the secondary relevance score. 
   
     
     
         15 . The method of  claim 14 :
 wherein updating the first cumulative relevance score based on the primary relevance score comprises summing the first cumulative relevance score and the primary relevance score; and   wherein updating the second cumulative relevance score based on the secondary relevance score comprises summing the second cumulative relevance score and the secondary relevance score.   
     
     
         16 . The method of  claim 1 , wherein updating the second edge connecting the first keyword node and the target video node based on the secondary relevance score comprises updating the second edge connecting the first keyword node and the target video node based on the secondary relevance score, the second edge representing the temporal position of the source video segment relative to the temporal position of the target video segment. 
     
     
         17 . A method comprising:
 accessing:
 a target node representing a target video segment in an audiovisual composition; and 
 a source video node representing a source video segment in the audiovisual composition; 
   accessing a first keyword associated with the source video segment based on a first edge connecting the source video node and a first keyword node representing the first keyword;   calculating a secondary relevance score for the first keyword relative to the target video segment based on:
 a temporal position of the source video segment in the audiovisual composition; and 
 a temporal position of the target video segment in the audiovisual composition; and 
 updating a second edge connecting the first keyword node and the target video node based on the secondary relevance score. 
   
     
     
         18 . The method of  claim 17 :
 further comprising:
 identifying the first keyword based on the source video segment; 
 calculating a primary relevance score for the first keyword relative to the source video segment; and 
 updating the first edge connecting the first keyword node and the source video node based on the primary relevance score; and 
   wherein calculating the secondary relevance score for the first keyword relative to the target video segment comprises calculating the secondary relevance score for the first keyword relative to the target video segment based on:
 the temporal position of the source video segment in the audiovisual composition; 
 the temporal position of the target video segment in the audiovisual composition; and 
 the primary relevance score for the first keyword relative to the source video segment. 
   
     
     
         19 . A method comprising:
 accessing:
 a target node representing a target video segment in an audiovisual composition; and 
 a source video node representing a source video segment in the audiovisual composition; 
   identifying a first keyword associated with the source video segment, the first keyword represented by a first keyword node;   calculating a primary relevance score for the first keyword relative to the source video segment;   updating a first edge connecting the first keyword node and the source video node based on the primary relevance score;   calculating a secondary relevance score for the first keyword relative to the target video segment based on:
 a temporal position of the source video segment in the audiovisual composition; 
 a temporal position of the target video segment in the audiovisual composition; and 
 the primary relevance score for the first keyword relative to the source video segment; and 
   updating a second edge connecting the first keyword node and the target video node based on the secondary relevance score.   
     
     
         20 . The method of  claim 19 , wherein calculating the secondary relevance score for the first keyword relative to the target video segment comprises:
 calculating an adjacency of the source video segment to the target video segment based on the temporal position of the source video segment in the audiovisual composition and the temporal position of the target video segment in the audiovisual composition; and   calculating the secondary relevance score for the first keyword based on the adjacency of the source video segment to the target video segment and the primary relevance score for the first keyword relative to the source video segment.