Systems and methods for machine learning predictions of the impact of digital content

Abstract

A system, computer readable medium, and method for analyzing digital content of electronic media files includes presenting control media content to a set of control respondents for consumption and presenting test media content to a set of test respondents for consumption. The method includes receiving first responses to a survey related to topics of the control media content from the set of control respondents and second responses to the survey about the test media content from the set of test respondents. The method includes performing feature extraction on the test media content and performing feature extraction on the first responses and the second responses. The feature extraction obtains response features associated with the first responses and the second responses. The method includes training a regression machine learning model with the media content features and the response features.

What is claimed is: 
     
         1 . A method for analyzing digital content of electronic media files, the method comprising:
 determining a set of control respondents and a set of test respondents;   presenting control media content to the set of control respondents for consumption;   presenting test media content to the set of test respondents for consumption;   receiving first responses to a survey about the test media content from the set of control respondents and second responses to the same survey about the test media content from the set of test respondents;   storing the first responses and the second responses, wherein the database stores the control media content and the test media content;   performing feature extraction on the test media content, wherein the feature extraction obtains media content features associated with the test media content;   performing feature extraction on the first responses and the second responses, wherein the feature extraction obtains response features associated with the first responses and the second responses; and   training a regression machine learning model with the media content features and the response features, wherein the model, when trained, outputs one or more of an importance indication for one or more of the media content features, a direction of influence for the one or more media content features, and an influence score of the test media content.